Thesis defenses in all programs should be announced online 10 days before the oral examination. The details of the thesis presentations are listed here and the calendar located in "Home" page. The students are strongly encouraged to attend the presentations.


Schedule


Speaker: Nazlı Deniz Türe

Affiliation PhD in Cryptography

Advisor: Assoc. Prof. Dr. Oğuz Yayla

Co-Advisor: Prof. Dr. Murat Cenk

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: 06.09.2024 – 15.00

Abstract: Digital signatures ensure authenticity and secure communication. They are used to verify the integrity and authenticity of signed documents and are widely utilized in various fields such as information technologies, finance, education, and law. They are crucial in securing servers against cyber attacks and authenticating connections between clients and servers. Additionally, encryption is used in many areas, such as secure communication, cloud, server and database security to ensure data confidentiality. Performing batch encryption, signature generation, and signature verification simultaneously and efficiently is highlighted as a beneficial approach for many systems. This work focuses on efficient batch signature generation with Dilithium, batch verifications of signatures from the same user using Crystals Dilithium (NIST's post-quantum digital signature standard) and batch encryption to a single user with Crystals Kyber (NIST's post-quantum encryption/KEM standard). One of the main operations of Dilithium and Kyber is the matrix-vector product with polynomial entries. So, the naive approach to generate/verify $m$ signatures with Dilithium (or encrypt $m$ messages with Kyber) where $m>1$ is to perform $m$ such multiplications.  In this paper, we propose to use efficient matrix multiplications of sizes greater than four to generate/verify $m$ signatures with Dilithium and greater than two to encrypt $m$ messages with Kyber. To this end, batch algorithms that transform the polynomial matrix-vector multiplication in Dilithium's and Kyber's structures into polynomial matrix-matrix multiplication are designed. The batch numbers and the sizes of the matrices to be multiplied based on the number of repetitions of Dilithium's signature algorithm are determined. Also, batch versions of Dilithium verification and Kyber encryption algorithms are proposed. Moreover, many efficient matrix-matrix multiplication algorithms, such as Strassen-like multiplications and commutative matrix multiplications, are analyzed to design the best algorithms that are compatible with the specified dimensions and yield improvements. Various multiplication formulas are derived for different security levels of Dilithium signature generation, verification, and Kyber encryption. Improvements up to 28.1%, 33.3%, and 31.5% in the arithmetic complexities are observed at three different security levels of Dilithium's signature, respectively. The proposed batch Dilithium signature algorithm and the efficient multiplication algorithms are also implemented, and 34.22%, 17.40%, and 10.15% improvements on CPU cycle counts for three security levels are obtained. The multiplication formulas used for batch Dilithium signature generation are also applied for batch Dilithium verification. At three different levels of security, improvements in the arithmetic complexity are observed of up to 28.13%, 33.33%, and 31.25%. Furthermore, 49.88%, 56.60%, and 61.08% improvements on CPU cycle counts for three security levels are achieved, respectively. As a result of implementing Kyber Batch Encryption with efficient multiplication algorithms, 12.50%, 22.22%, and 28.13% improvements on arithmetic complexity, as well as 22.34%, 24.07%, and 30.83% improvements on CPU cycle counts, are observed for three security levels.


Speaker: Gizem Damla Ateşağaoğlu Alan

Affiliation MSc in Financial Mathematics

Advisor: Prof. Dr. Ali Devin Sezer

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: 05.09.2024 / 10.00-10:30

Abstract: : A basic market making model in mathematical finance is the one proposed by Avellaneda and Stoikov (AS), which is formulated as a stochastic optimal control problem. This model includes a risk liquidity premium function ℓ which penalizes the remaining inventory at terminal time. To the best of our understanding, in the currently available literature, at least in the context of the AS model, this function is usually assumed zero or is ignored. One explanation given for this assumption is that the dynamics of the inventory process is mean reverting and the ℓ function has little impact on this. In this thesis, we study numerically whether this assumption holds by computing the dynamics of the inventory process for non-zero ℓ functions. We see that depending on model parameters this function can have a nontrivial impact on inventory process dynamics. We present a numerical study of how this impact depends on model parameters. 


Speaker: Firdevs Nur Uykun

Affiliation MSc in Financial Mathematics

Advisor: Assist. Prof. Dr. Büşra Z. Temoçin

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: 05.09.2024 / 11.00-11:30

Abstract: This study examines the evaluation of S&P 500 trends and their effects on method ologies used in portfolio optimization, with a particular focus on achieving diversifi cation among 24 stocks representing eight different industry sectors and one risk-free asset. Through the meticulous construction of risk aversion-adjusted portfolios appli cable to both single and multiple period analyses, the research employs variance and Conditional Value at Risk (CVaR) for single periods, while using Mean Absolute De viation (MAD) for multiple periods. The portfolio optimization process benefits from the synergistic use of Python for computational modeling and AMPL (A Mathemat ical Programming Language) for executing complex mathematical formulations. To accurately predict risk aversion behaviors, the study utilizes six classification mod els integrated with 29 indicators derived from technical analysis: Logistic Regression (LR), K-Nearest Neighbors (KNN), Support Vector Machines (SVM), Decision Trees (DT), Random Forest (RF), and eXtreme Gradient Boosting (XGBoost). Simultane ously, return forecasting leverages the predictive capabilities of four regression frame works—Linear Regression, Long Short-Term Memory (LSTM), XGBoost, and Light Gradient Boosting Machine (LightGBM)—based on various technical indicators. Ex plainable Artificial Intelligence (XAI) techniques, particularly Local Interpretable Model-agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP), facilitate a deeper understanding of feature importance in the decision-making pro cesses of machine learning algorithms. The findings of this research show that the integration of machine learning algorithms has the potential to significantly enhance portfolio management strategies, thereby providing innovative insights in the pursuit of optimized asset allocation and risk management. 


Speaker: Melis Aslan

Affiliation PhD  in Cryptography

Advisor: Assoc. Prof. Dr. Ali Doğanaksoy

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: 02.09.2024 – 14.15

Abstract: Random numbers play a crucial role in cryptography, since security of cryptographic protocols relies on the assumption of availability of uniformly distributed and unpredictable random numbers to generate secret keys, passwords, initialization vectors, nonces, salt, etc. True Random Number Generators (TRNGs) extract random numbers from physical processes (such as radioactive decay, thermal noise, and atmospheric noise) that are inherently unpredictable. However, real-world random number generators sometimes fail and produce outputs with low entropy, leading to security vulnerabilities. It is commonly observed that the TRNG outputs have statistical biases and dependencies that make them unsuitable to be directly used for cryptographic purposes. There are some standards and guidelines on generating and testing random numbers that are suitable for cryptographic applications. The National Institute of Standards and Technology (NIST) Special Publication (SP) 800-90 series provide guidelines and recommendations for generating random numbers for cryptographic applications and describes statistical randomness testing, estimating min-entropy with 10 black-box entropy estimation methods. In this thesis, we evaluate the effectiveness and limitations of the SP 800-90B methods by exploring the accuracy of these estimators using simulated random numbers with known entropy, investigating the correlation between entropy estimates, and studying the impacts of deterministic transformations on the estimators.To understand the unpredictability of the outputs, it is important to estimate their entropy accurately. The National Institute of Standards and Technology (NIST) Special Publication (SP) 800-90B specifies ten min-entropy estimators ranging from simple frequency-based estimators to more advanced approaches. Each estimator has specific assumptions, making them suitable for different types of sources. The minimum of these estimates is assumed to be the min-entropy of the TRNG. We propose a new entropy estimator (estimates min-entropy and also  Shannon entropy) called index-value coincidence estimate that is suitable for outputs that might include some dependencies and we also provide some experimental results that demonstrate the effectiveness of the estimator.Additionally, TRNGs may be affected by outside conditions such as temperature, humidity etc. Health tests are an integral part of the noise source of TRNG defined to detect unexpected changes in the working process and dramatic changes in the amount of the entropy that generated by the noise source. Existing health test suites are examined and a health test suite for cryptographic TRNGs is introduced by using random variables weight, run, runs of length 1 and overlapping templates, some suggested parameters and experimental results are given.


Speaker: Emine Ezgi Alptekin

Affiliation : PhD in Financial Mathematics

Advisor: Prof. Dr. Ömür Uğur

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: 02.09.2024 / 10.30

Abstract: In many fields like physics, ecology, biology, economics, engineering, and financial mathematics, events often don't have an immediate effect. Instead, they impact future situations. To understand how these systems work and behave, we use stochastic delay differential equations (SDDEs). These equations include information from past events into stochastic differential equations (SDEs). SDDEs are gaining attention because they can better reflect real-life situations.Some numerical methods for SDDEs have been developed because it's often very difficult, and sometimes impossible, to find exact solutions using stochastic calculus. Recently, researchers in economics and finance have been studying option pricing for systems with time delays, which can be either random or fixed. This thesis aims to understand the general forms of SDDEs and the process for solving them when the time delay is fixed especially for the European type options.


Speaker: Türker Açıkgöz

Affiliation MSc Financial Mathematics

Advisor: Assist. Prof. Dr. Büşra Z. Temoçin

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: 27.08.2024 / 09:30-10:00

Abstract:  The Efficient Market Hypothesis has been dominating the literature of finance for a long time. Meanwhile, the problematic assumptions and inappropriateness of Efficient Market Hypothesis in explaining real-life financial markets have dictated the significance of developing new theories and approaches. Up to now, among the literature on finance, the Fractal Market Hypothesis has developed as an alternative to the Efficient Market Hypothesis. The Fractal Market Hypothesis is developed based on fractal geometry and fractal Brownian motions, and their applications on financial markets. In essence, the hypothesis postulates that financial markets are structured as fractals, they exhibit statistical self-similarity, and long-term memory in their time series. On the other hand, the various portfolio applications to this hypothesis are quite limited in the literature. The present portfolio studies have a number of basic problems, including the definition of covariance, inability to propose a portfolio for more than two assets, and detrending. In this thesis, a portfolio optimization approach based on Fractal Market Hypothesis is developed which takes into account these problems of the existing models in the literature. This thesis proposes a portfolio optimization method, the Mean-MFTWXDFA (Mean-Multifractal Detrended Temporally Weighted Detrended Cross-Correlation Analysis) which is based on multifractal temporally weighted cross-correlation analysis with detrending approach by geographically weighted regression. The suggested method is also compared with those of classical portfolio applications such as the Mean-Variance, Mean-Value at Risk, and Mean-Conditional Value at Risk methods. The portfolio analysis include cryptocurrency market and three diversifying assets: oil, clean energy and equity. Applications of the fractal-based portfolio do reasonably well into out-of-sample analyses and outperform conventional ones. 


Speaker: İlkyaz Aslanöz

Affiliation MSc in Actuarial Sciences

Advisor: Prof. Dr. A. Sevtap Selçuk-Kestel

Co-Advisor: Dr. Bükre Yıldırım Külekci

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: 31.07.2024 / 9.30-11.00

Abstract:  In non-life insurance, the accurate estimation of total loss is extremely important for companies' asset-liability management. To estimate the total loss, insurance companies use generalized linear model (GLM) as it is compatible with insurance data and hence makes considerably consistent predictions. The common practice is constructing a GLM for frequency, usually using the Poisson distribution, and another GLM for severity, usually using the Gamma distribution, and then multiplying the results of these two models. However, this multiplication is only possible under the assumption of independence. Although this assumption simplifies the modeling, it also causes deviations from the real loss value. In this thesis, constructing a GLM, which can incorporate the dependence between claim frequency and severity, to predict the total loss is aimed. Two GLMs are built and tested; dependent-GLM and copula-GLM, and they are compared with the regular independent-GLM. To examine these models, non-life motor third party liability (MTPL) insurance data is used. The data is split into two parts; 80\% of it is used to construct models, and 20\% of it is used to test these models. In the first model, the dependency is provided by taking the claim number as a covariate of marginal severity GLM. The second model provides the dependence between the marginal frequency and severity GLMs by using a copula function. To compare these three models, Akaike Information Criterion (AIC) is used, and also the means of the estimations of 


Speaker: Anıl Gülveren

Affiliation PhD in Financial Mathematics

Advisor: Prof. Dr. A. Sevtap Selçuk-Kestel

Co-Advisor: Assist. Prof. Dr. Başak Bulut Karageyik

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: 31 July 11-12:30 

Abstract:  This thesis investigates optimal investment and reinsurance strategies aimed at maximizing the expected utility of wealth. Building on foundational work in optimal investment and insurance strategy within a diffusion setup, we extend the model by incorporating an Ornstein-Uhlenbeck process to model the instantaneous return rate of the risky asset. This approach provides a more detailed and realistic representation of market fluctuations, capturing both bull and bear markets. Our framework allows short selling, aligning it more closely with real market practices where investors use diverse strategies to manage risk and enhance returns. We develop the related Hamilton-Jacobi-Bellman (HJB) equation and derive closed-form expressions for the optimal investment and reinsurance decisions. In addition to developing the theoretical model, we conduct extensive numerical simulations to illustrate how different financial and insurance parameters influence the optimal strategies. The sensitivity analysis provides critical insights into how variations in parameters such as volatility, risk aversion, and time horizon impact the insurer’s decisions. Specifically, we find that the optimal investment strategy tends to increase in bullish markets and decrease with higher volatility, reflecting the insurer’s need to balance risk and return. The optimal reinsurance strategy exhibits threshold behavior, remaining zero until a certain level of risk aversion is reached, beyond which it increases significantly. By incorporating realistic market dynamics and providing actionable insights, these advancements contribute to more effective risk management and strategic planning in volatile markets, providing a more robust and flexible framework for decision-making.


Speaker: Yunus Emre Yılmaz

Affiliation PhD - Cryptography Program

Advisor: Assoc. Prof. Dr. Oğuz YAYLA

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: 26.07.2024 / 12:00-13:00 

Abstract:  True Random Number Generators (TRNGs) and Physically Unclonable Functions (PUFs) are two basic and useful primitives in designing cryptographic systems. TRNGs must be invariably random, while PUFs must have repetitive results and instance-specific randomness. TRNGs are designed to generate random padding bits, nonces, and initialization vectors, whereas PUFs are designed to extract chip-unique signatures and volatile secret keys. In order to implement these primitives in hardware, application-specific integrated circuits (ASICs) or Field-Programmable Gate Array (FPGA) can be chosen. Although, ASICs may perform better than FPGAs, once they are designed, it is not easy or practical to change the silicon design. However, FPGAs offer easily changeable configurations for hardware implementation. Recently popular System-on-Chip (SoC) FPGAs or simply SoCs are semiconductor devices that integrate programmable logic with hard processor cores. They offer higher integration, lower power, smaller board size, and higher bandwidth communication between the processor and FPGA. Both in FPGA and in SoCs, Phase-Locked Loops (PLLs) are required and placed. Briefly, those are a feedback control system that automatically adjusts the phase of a locally generated signal to match the phase of an input signal. This PLL structures can be used to design TRNGs [2]. This PLL structure and internal auxiliary delay circuits in a FPGA can be used to implement a PUF unit on a FPGA [6]. Combining TRNG and PUF on a single device is important, because a root-of-trust for an embedded device can be implemented by these two PUF together. In thesis work, an efficient and integrated TRNG and PUF hardware design on an SoC is presented. The results of TRNG are examined with respect to BSI criteria using AIS20/31 [5], while the results of PUF are analyzed using [4].


Speaker: Nurşen Karasu

Affiliation MSc in Financial Mathematics

Advisor: Prof. Dr. Erkan Erdil

Co-Advisor: Prof. Dr. Kasırga Yıldırak

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: 03.07.2024 / 10:30

Abstract:  Climate conditions have a big impact on the yield of farm products, hence the prices. This thesis makes price prediction of majorly traded grains, Wheat, Barley and Corn, based on major climate conditions, total precipitation and dew point, levels of which are taken from European Centre for Medium-Range Weather Forecast (ECMWF) database of Konya, Polatli, Yozgat, Adana and Urfa where major production is held, by using a smart, machine learning model of ensemble training and compares with the price prediction results retrieved by simple regression method. These predictions would help future prices in futures commodity markets to be determined and any desired future price index can then be retrieved through these settled end of day future prices, by applying weights decided by index structurers. The study also confirms that tree-fit model could be used for future price prediction even when the statistic significance of data is low, therefore a successful simple regression optimization methodology is hard to apply. On the other hand, high volatile inflation rate and exchange rate would lead the predictions to deviate outside the accepted limits so the model studied in this paper is believed to be a better fit in stabilized economies. This thesis study offers important clues for producers for their production preferences, policy makers to build production planning, insurance companies to make climate risk mappings and financial insrument traders to make reasonable pricing so to prevent volatility in commodity market. Finally, all thesis study outputs are expected to serve the purpose of maintaining a sustainable agricultural production.


Speaker: Sena Kahveci

Affiliation MSc in Programme Financial Mathematics

Advisor: Prof. Dr. Z. Nuray Güner

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 26 April 2024; 13:30

Abstract:  "In recent years, environmental factors seem to play important roles in financial decisions, particularly due to the increasing popularity of sustainable investing strategies. An important investment alternative in this field is green commercial mortgage-backed securities (CMBS), complex financial products supported by loans on commercial properties with environmental certifications or sustainable characteristics. This study investigates the presence of a "greenium", defined as a pricing premium or discount associated with environmentally sustainable attributes of assets, for green CMBS using a large dataset of green and traditional CMBS. To our knowledge, this issue has not been examined in the context of commercial mortgages and commercial mortgage backed securities. In this thesis, the existence of greenium at the time of issuance in notes rates for mortgages on green-labeled and traditional multifamily properties and greenium in the pass-through rates for green and traditional CMBS are analyzed using OLS regressions while controlling for property, mortgage, and market characteristics. Results indicate that green-labeled multifamily properties are financed with mortgages having note rates that are 9.4 basis points lower than mortgages on comparable non-green properties, hence there is greenium in the mortgage notes rates.  However, when these mortgages are securitized, the pass-through rate for only CMBS with green certification exhibits 13.77 basis point greenium. These results are consistent with the existence of greenium in the primary mortgage and CMBS markets. When the differences in pricing of green and traditional CMBS in the secondary market are analyzed using the Markowitz mean-variance optimization technique, mixed results are obtained for the existence of greenium. The findings of thesis add to the expanding knowledge of sustainable finance, and the pricing of commercial mortgages and CMBS. Furthermore, the findings have significant consequences for investors, policymakers, and market participants interested in comprehending the connection between environmental conditions and the valuation of commercial mortgage-backed securities."


Speaker: Alper Umut Uçar

Affiliation MSc

Advisor: Assoc. Prof. Dr.Oğuz Yayla

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 24 April 2024; 12:30

Abstract:  This thesis presents a comprehensive study on the integration of ML and advanced privacy-preserving technologies in the rapidly evolving field of vehicular networks, particularly in the context of emerging 6G telecommunications. We explore vari- ous ML algorithms, including supervised, unsupervised, semi-supervised, and rein- forcement learning, and their applicability in vehicular networks for enhancing safety, navigation, and traffic management. Special emphasis is placed on the critical need for privacy preservation in this highly interconnected domain. The thesis categorizes privacy-preserving technologies into soft and hard models, analyzing their roles and impacts within vehicular networks. Soft privacy models, reliant on third-party enti- ties, are contrasted with hard privacy models that emphasize data minimization and limit third-party data disclosure. The implications of these models are examined in the context of vehicular networks, considering the latest advancements in DLT and blockchain, which are instrumental in enhancing security and operational efficiency. This thesis aims to provide a balanced perspective on the trade-offs between privacy, system capabilities, and cost in vehicular networks, offering insights into the future direction of these technologies in the era of 6G.


Speaker: Şeref Kutay Yakut

Affiliation MSc in Financial Mathematics

Advisor: Prof. Dr. Ali Devin Sezer

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 16 April 2024; 13:30

Abstract:  One of the first works estimating jump risk premium in financial markets is the seminal work of Jun Pan published in 2000. In this work Pan uses the generalized method of moments (GMM) to estimate the parameters of a stochastic volatility price model with jumps from index and option price data.  In the implementation of GMM, Pan uses a set of optimal moment conditions. In this thesis, we simulate the stochastic model used in Pan's work and apply the GMM estimation algorithm using ordinary moment conditions on simulated data. The estimation results suggest that the ordinary moment conditions are not very sensitive to model parameters and as a result the estimation algorithm quickly converges to a point around the initial parameter estimate.  We applied the same algorithm to a stock price and a call option observed in Borsa İstanbul and observed a similar performance.


Speaker: Gökhan Elmas

Affiliation MSc in Cryptography

Advisor: Assist. Prof. Dr. Buket Özkaya

Co-Advisor Prof. Dr.  Ferruh Özbudak

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 26 January 2024; 18:00

Abstract:  : Mutually unbiased bases (MUB) defined for Hilbert spaces are a specific type of bases with a constraint stated in terms of inner products. Their constructions on different dimensions lead to some open problems besides their construction methods. In this thesis study, we explore some ways to construct MUBS over Hilbert spaces of dimension of prime power. We also introduce the multidisciplinary background needed for the corresponding constructions.


Speaker: Sermin Kocaman

Affiliation PhD  in Cryptography

Advisor : Assoc.Prof. Dr. Ali Doğanaksoy

Co-Advisor :  Assoc.Prof. Dr.  Fatih Sulak

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 26 January 2024; 15:00

Abstract:  Helios is the first web-based and open-audit voting system. The open-audit feature allows anyone to track the voting process, thus providing easy verifiability in all stages of the elections. Despite many advantages, Helios has a few weaknesses due to its reliance on a centralized server, such as modifying data through unauthorized access or making the server inaccessible. A subsequent work called blockchain-powered Helios is proposed to overcome these weaknesses. This system replaced the centralized server with decentralized servers using the blockchain and a decentralized storage protocol. Although this novelty eliminates Helios' centralized weaknesses,  it creates some new problems, thereby causing security weaknesses. These are the misbehavior in the wallet registration procedure, the linkability in the wallet authorization procedure, and the high cost of transactions. In this thesis, an improved version of the blockchain-powered Helios system, named Proba, is presented. The system is redesigned to provide privacy-preserving, robustness, and accessibility in the election. Proba utilizes a novel threshold issuance-anonymous credential that breaks the link between the voters and their wallets. Also, the threshold version mitigates the misbehavior of election authorities in wallet registration. In addition to these, Proba utilizes consortium blockchain that provides affordable election costs. The performance analysis of Proba shows that the usage of threshold-issuance anonymous credentials does not have a critical cost for the election phase; on the contrary, it mitigates the smart contract storage cost.

 As an additional work on the design of Proba, this thesis contributes to the protection of the voter's wallet signing key by accelerating the two-party elliptic curve digital signature algorithm. Within the blockchain, transactions must be signed using the wallet's signing key. Thus, the voter's right to send transactions on the blockchain will depend on the security of this single key. To protect the wallet signing key, the voter may choose to store this key in two distinct locations, such as a smartphone and a laptop. In this scenario, the attacker must simultaneously acquire access to both locations, but this is not a realistic scenario. Thus, in this thesis, it is recommended that voters split wallet signing keys into two separate locations to protect the keys using a two-party signing mechanism that reduces bandwidth usage. This protocol offers the most efficient offline phase for a two-party ECDSA protocol with such an efficient online phase.

Furthermore, this thesis enables the utilization of the voter's signing keys in a hierarchical structure for key protection. Given that phones are considered important devices for verifying an individual's identity, the voter has the option to divide their signing key across their phone and other devices. In this scenario, it is mandatory to use the phone, which is considered to be in the highest hierarchy, during the signing phase. However, signing keys shared with other devices can be used according to the specified threshold value in the system. Existing hierarchical threshold schemes contain certain ordering rules and constraints at each hierarchical level, limiting their adaptability. However, the proposed  FlexHi scheme offers a new architecture that breaks free from these constraints and offers flexibility.


Speaker: Bilge Nur Mermer

Affiliation MSc in Actuarial Sciences

Advisor: Assist. Prof. Dr.  Büşra Z. Temoçin

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 26 January 2024; 11:00

Abstract: The new type of coronavirus disease, which emerged for the first time in December 2019, has caused a significant number of cases and deaths worldwide up to the present day. In addition, during the week when the epidemic was declared as a pandemic, US crude oil prices declined to their lowest level in the last 18 years. It is important to model the transmission dynamics of the virus and predict its development. Although there are multiple methods in modeling the spread of the virus, the most known and most used one is the deterministic SIR model. In this thesis, various stochastic extensions of SIR model are used and compared with the deterministic one. The basic reproduction number Ro to be obtained from of the model, expresses how many people an infected person will infect, and shows to what extent the epidemic is brought under control. This thesis aims to develop a prediction grid based on the basic reproduction number and enhances comparisons of models with the basis of deterministic SIR model. The effectiveness of the results is examined by forecasting Ro based on US. The findings of this research are expected to form an important basis for making policy comparisons for many industries, such as health, finance and governance. In addition, with the Covid-19 pandemic process, the direction, intervals and duration of fluctuations, especially in the US financial markets are examined due to the data availability and its influence on the world’s financial markets.


Speaker: Anıl Şen

Affiliation MSc in Programme Actuarial Sciences

Advisor: Prof. Dr. A. Sevtap Kestel

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 26 January 2024; 10:00

Abstract:  This thesis explores how to improve the way insurance companies estimate losses, focusing on two methods: the Chain-Ladder technique and ARIMA forecasting. It highlights the importance of short-term estimating losses, especially in a high inflation environment. The study emphasizes the adaptability and responsiveness of short-term loss estimation. By integrating ARIMA, a method for analyzing time series data, the research aims to provide a systematic approach for accurate forecasting in the insurance data. The results aim to improve the accuracy and reliability of loss estimation, emphasizing the importance of short-term estimates in dealing with uncertainties. This research aims to offer practical insights for those working in the insurance industry, helping them make more effective decisions in this constantly changing field.


Speaker: Yunus Emre Yılmaz

Affiliation PhD - Cryptography Program

Advisor: Assoc. Prof. Dr. Oğuz Yayla

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 25 January 2024; 13:00

Abstract: True Random Number Generators (TRNGs) and Physically Unclonable Functions (PUFs) are two basic and useful primitives in designing cryptographic systems. TRNGs must be invariably random, while PUFs must have repetitive results and instance-specific randomness. TRNGs are designed to generate random padding bits, nonces, and initialization vectors, whereas PUFs are designed to extract chip-unique signatures and volatile secret keys. In order to implement these primitives in hardware, application-specific integrated circuits (ASICs) or Field-Programmable Gate Array (FPGA) can be chosen. Although, ASICs may perform better than FPGAs, once they are designed, it is not easy or practical to change the silicon design. However, FPGAs offer easily changeable configurations for hardware implementation. Recently popular System-on-Chip (SoC) FPGAs or simply SoCs are semiconductor devices that integrate programmable logic with hard processor cores. They offer higher integration, lower power, smaller board size, and higher bandwidth communication between the processor and FPGA. Both in FPGA and in SoCs, Phase-Locked Loops (PLLs) are required and placed. Briefly, those are a feedback control system that automatically adjusts the phase of a locally generated signal to match the phase of an input signal. This PLL structures can be used to design TRNGs [2]. This PLL structure and internal auxiliary delay circuits in a FPGA can be used to implement a PUF unit on a FPGA [6]. Combining TRNG and PUF on a single device is important, because a root-of-trust for an embedded device can be implemented by these two PUF together. In thesis work, an efficient and integrated TRNG and PUF hardware design on an SoC is presented. The results of TRNG are examined with respect to BSI criteria using AIS20/31 [5], while the results of PUF are analyzed using [4].

[2] V. Fischer and M. Drutarovský, True Random Number Generator Embedded in Reconfigurable Hardware, in B. S. Kaliski, Ç. K. Koç, and C. Paar, editors, Cryptographic Hardware and Embedded Systems - CHES 2002, pp. 415–430, Springer Berlin Heidelberg, Berlin, Heidelberg, 2003, ISBN 978-3-540-36400-9.

[4] A. Maiti, V. Gunreddy, and P. Schaumont, A Systematic Method to Evaluate and Compare the Performance of Physical Unclonable Functions, Cryptology ePrint Archive, Paper 2011/657, 2011, https://eprint.iacr.org/2011/657.

[5] M. Peter and W. Schindler, A Proposal for Functionality Classes for Random Number Generators - Version 2.35 - DRAFT, 2022, https://www.bsi.bund.de/SharedDocs/Downloads/EN/BSI/Certification/Interpretations/AIS_31_Functionality_classes_for_random_number_generators_e.pdf?__blob=publicationFile&v=7.

[6] K. Pratihar, U. Chatterjee, M. Alam, R. S. Chakraborty, and D. Mukhopadhyay, Birds of the Same Feather Flock Together: A Dual-Mode Circuit Candidate for Strong PUF-TRNG Functionalities, IEEE Transactions on Computers, 72(6), pp. 1636–1651, 2023.


Speaker: Alper Umut Uçar

Affiliation MSc in Cryptography

Advisor: Assoc. Prof. Dr. Oğuz YAYLA

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 22 January 2024; 13:00

Abstract: This thesis presents a comprehensive study on the integration of ML and advanced privacy-preserving technologies in the rapidly evolving field of vehicular networks, particularly in the context of emerging 6G telecommunications. We explore vari- ous ML algorithms, including supervised, unsupervised, semi-supervised, and rein- forcement learning, and their applicability in vehicular networks for enhancing safety, navigation, and traffic management. Special emphasis is placed on the critical need for privacy preservation in this highly interconnected domain. The thesis categorizes privacy-preserving technologies into soft and hard models, analyzing their roles and impacts within vehicular networks. Soft privacy models, reliant on third-party enti- ties, are contrasted with hard privacy models that emphasize data minimization and limit third-party data disclosure. The implications of these models are examined in the context of vehicular networks, considering the latest advancements in DLT and blockchain, which are instrumental in enhancing security and operational efficiency. This thesis aims to provide a balanced perspective on the trade-offs between privacy, system capabilities, and cost in vehicular networks, offering insights into the future direction of these technologies in the era of 6G.


Speaker: Ahmet Ramazan Ağırtaş

Affiliation PhD in Cryptography

Advisor: Assoc. Prof. Dr. Oğuz YAYLA

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 22 January 2024; 10:00

Abstract: An accountable subgroup multi-signature (ASM) is a kind of multi-signature scheme in which any subgroup S of a group G of potential signers jointly sign a message m, ensuring that each member of S is accountable for the resulting signature. In this thesis, we contribute to the literature in three directions. In the  first direction, we propose three novel pairing-based ASM schemes (vASM, ASMwSA, and ASMwCA), which are secure against existential forgery under chosen-message attacks and computational co-Die-Hellman assumption. In the  first one (vASM), we use Feldman's verifiable secret sharing scheme as an implicit authentication and proof-of-possession for setting up group G. In the second one (ASMwSA), the members participating in authentication are decided by the subgroup. In the third one (ASMwCA), we  consider a designated combiner managing the authentication process. All pairing-based schemes we propose require fewer computations in the signature generation, signature aggregation, and verification phases than the pairing-based ASM scheme proposed by Boneh, Drijvers and Neven. More over, our first and third ones (vASM and ASMwCA) solve the open problem of constructing an ASM scheme in which the subgroup S of signers is unknown before the signature generation. Besides, we give a method of eliminating the combiner in case of knowing the subgroup of signers S in advance. Further, we extend our proposed schemes to aggregated versions. For N many ASM signatures, aggregated versions of our proposed pairing-based schemes output an aggregated signature with the size of a single group (G1) element and require N + 1 pairings in aggregated signature verification. In contrast, the partially aggregated ASM scheme of Boneh, Drijvers and Neven gives an aggregated signature with the size of N + 1 group elements and requires 2N + 1 pairings in aggregated signature verification. In the second direction, we study the compartment-based and hierarchical threshold delegation of signing power of the verifiable accountable subgroup multi-signature (vASM) scheme. We show that the vASM scheme can also be considered as a proxy signature in which an authorized user (original signer, designator) delegates her signing rights to a single (or a group of) unauthorized user(s) (proxy signer). Namely, wefipropose four new constructions with the properties and functionalities of an ideal proxy signature and a compartment-based/hierarchical structure. In the  first construction, we apply the vASM scheme recursively; in the second one, we use Shamir's secret sharing (SSS) scheme; in the third construction, we use SSS again but in a nested fashion; and in the last one, we use the hierarchical threshold secret sharing (HTSS) scheme for delegation. Then, we show the a liation of our constructions to proxy signatures and compare our constructions with each other in terms of efficiency and security. Finally we compare the vASM scheme with the existing pairing-based proxy signature schemes. In the third direction, we propose 1 a novel lattice-based ASM scheme, i.e. vMS2, by combining the group setup method of vASM scheme and Damgard et al.'s lattice-based MS2 multi-signature scheme. Key generation, signature generation and verification phases of our proposed scheme are almost identical to the MS2 scheme. In the group setup phase, we generate membership keys which is used for signing a message on behalf of a group G of users. These membership keys are generated via a joint verifiable secret sharing (VSS) scheme in a way that they include a piece of information from the secret keys of all users in G so that any subgroup of users in G having a valid membership key can sign in an accountable fashion. We also present a comparison of the underlying MS2 scheme and our accountable subgroup multi-signature scheme vMS2 to show the cost of accountability. We show that lattice-based ASM scheme can be achieved by adding a one-time one-round group setup whose cost is slightly higher than signature generation and verification of theunderlying MS2 signature scheme.


Speaker: Şeref Kutay Yakut

Affiliation MSc in Financial Mathematics

Advisor: Prof. Dr. Ali Devin Sezer

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 17 January 2024; 11:00

Abstract: Following the seminal work of Pan on the estimation of jump-risk premia, we fit a continuous time stochastic differential equation with jumps model to a stock price traded in Borsa Istanbul and an option price on the same stock. The SDE model includes a stochastic volatility component in addition to the jumps. From this fit we infer a jump premium in the excess return offered by the stock. This premium is computed for two different months in 2023 to see how the premium changed in these two periods.


Speaker: Korkut Anapa

Affiliation MSc in Scientific Computing

Advisor: Assoc. Prof. Dr. Hamdullah Yücel

Co-Advisor: Dr. Songül Bayraktar

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 15 December 2023; 12:00

Abstract: This thesis proposes a novel approach based on statistical learning and multi-objective optimization to reduce the need for experiments during the design phase of new cleaning cycles for household dishwashers. First, regression models are built that are associated with the feature selection methods to predict the outputs of a dishwasher cleaning cycle by using the existing cleaning cycles’ program flows as input data and the results of the performance laboratory tests of the related cleaning cycles as output data. Then, a multi-objective optimization problem is defined by assigning the regression models and chosen features as objective functions and unknown decision variables, respectively. The obtained optimization problem is then solved using evolutionary algorithms according to the designer’s preferences (or customers’ needs).


Speaker: Halil Ömer Tekin

Affiliation MSc in Cryptography

Advisor: Doç. Dr. Ali Doğanaksoy

Place: Room S211 Hall, IAM, METU

Date/Time: 11 December 2023; 15:00

Abstract: For more than a century sorting algorithms have been subject of research. Merge Sort Algorithm, one of the earliest sorting algorithms still used, has recursive and nonrecursive implementations. Non-recursive (bottom-up) version runs slower than recursive one due to some issues including cache efficiency. However Bottom-Up version can be preferred to recursive one, for recursion avoided systems like aircraft, space programs. Recursion can also be prevented while working with encrypted data and encrypted indices because of the initialization or termination conditions. In this thesis, a non-recursive merge sort algorithm, called Omerge Sort, is proposed and analyzed to beat Bottom-Up Merge Sort. Furthermore, via using Hoare Logic, the formal proof is given that the algorithm actually sorts for the array lengths of power of 2.


Speaker: Can Deniz Çam

Affiliation MSc in Scientific Computing

Advisor: Assoc. Prof. Dr.  Esma Kaygısız

Co-Advisor: Assoc. Prof. Dr. Hamdullah Yücel

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 11 December 2023; 10:30

Abstract: All around us, we can observe individuals making relations with others to form networks. However, how these networks came to be and whether they are stable or not are more complex questions. Network formation games and associated literature is interested in finding how individuals may alter the networks they are in and describing their incentives to do so. This study aims to contribute to both of these endeavours. Firstly, the multiple connectivity approach, a way to compute how individuals may alter all their connections while still making short-term profit-maximising choices, is presented. Secondly, the Component Profit Model is proposed to describe how a firm may profit by lowering its production costs via network interactions. Both of these concepts are united in the Multiple Connectivity Game, where each player can propose an alteration to their networks, and a network is stable if no player is incentivised to do so.

The profit attainable by altering all possible connections simultaneously generates a complex optimisation problem. In order to reduce the complexity of such problems, this thesis proposes converting each connection choice to a discrete decision variable and converting the problem to a mixed integer optimisation problem. Moreover, a method to solve this problem using a modification of the branch and bound algorithm is discussed. Applying the proposed method requires treating discrete network properties as continuous values. An interpolation method that is particular to the problem is provided to facilitate the solvability of the problem. Finally, the ability of the proposed model to describe network formation is discussed by simulating various systems with different parameters.


Speaker: Fatih Aykurt

Affiliation MSc in Cryptography

Advisor: Assoc. Prof. Dr. Oğuz YAYLA

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Date/Time: : 07 December 2023; 14:00

Abstract: Using secure multi-party computing protocols (MPC), a group of participants who distrust one another can securely compute any function of their shared secret inputs. Participants exchange these inputs in a manner similar to secret sharing, where each participant owns a portion of the input but is unable to independently reconstruct the complete information without collaborating with the other participants. This kind of computation is quite powerful and has many uses where data privacy is quite critical such as areas like government, business, and academia. MPC has grown from a subject of theoretical study to a technology being employed in industry, becoming effective enough to be deployed in practice with various algorithms implemented with MPC frameworks. In this work, two versatile MPC frameworks, MP-SPDZ and MPyC are studied. These frameworks performances are compared by using execution times and profiling results are also analyzed from basic operations to more complex structures like shuffle sort algorithm. The bottleneck points of the algorithms where the time consumption increases drastically are revealed. To detect the critical parts easier, profiling results are visualized as dot graphs. In the MPyC framework, Sattolo shuffle algorithm is implemented and compared with the current modern version of Fisher-Yates algorithm.


Speaker: Halil Ömer Tekin

Affiliation MSc in Cryptography

Advisor: Doç. Dr. Ali Doğanaksoy

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 11 September 2023; 18:00

Abstract: For more than a century sorting algorithms have been subject of research. Merge Sort Algorithm, one of the earliest sorting algorithms still used, has recursive and nonrecursive implementations. Non-recursive (bottom-up) version runs slower than recursive one due to some issues including cache efficiency. However Bottom-Up version can be preferred to recursive one, for recursion avoided systems like aircraft, space programs. Recursion can also be prevented while working with encrypted data and encrypted indices because of the initialization or termination conditions. In this thesis, a non-recursive merge sort algorithm, called Omerge Sort, is proposed and analyzed to beat Bottom-Up Merge Sort. Furthermore, via using Hoare Logic, the formal proof is given that the algorithm actually sorts for the array lengths of power of 2.


Speaker: Ezgi Naz TEKİN

Affiliation MSc in Cryptography

Advisor: Prof. Dr. Ferruh ÖZBUDAK

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 11 September 2023; 15:30

Abstract: Homomorphic encryption (HE), which enables computations on encrypted data without first decrypting it, is a ground-breaking advancement in the cryptographic area. In this study, many HE schemes such as partially, somewhat, and fully are examined and the details of the algorithms in these methods are given. Furthermore, this comprehensive examination goes beyond theoretical considerations to include real-world applications of this encryption technique. It examines both the potential and the challenges of using HE as a powerful tool for ensuring security and privacy in current information systems. This exploration takes into account not only the theoretical foundations but also the practical aspects of its implementation.


Speaker: Dursun Oylum SERİNER GERENLİ

Affiliation MSc in Cryptography

Advisor: Prof. Dr. Ferruh ÖZBUDAK

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 11 September 2023; 15:00

Abstract: Neural networks are widely used learning models to achieve successful results in many application areas today. However, proving and sharing the accuracy and reliability of these networks is often limited due to privacy and security challenges. In this study, a method of cryptographic proving the accuracy of neural networks without revealing their intrinsic components is presented. The method is presented by using the Circom programming language to create a circuit containing these elements by making use of the final weights, bias values, and inputs of the neural networks. The use of the Circom programming language makes it possible to convert neural network elements into electronic circuits. The resulting circuit contains the representation of the neural network model and mimics the transformation from inputs to outputs. It is also used with Groth16 which is Zero Knowledge Proof system to prove the accuracy of the neural network without leaking private information. As in this study, the newly produced circuit can be used with the help of zkREPL or terminal. As a result, a method is presented to prove the real-world performance of the neural network model and increase the reliability of the model. In this way, the correctness of the model can be proven without directly telling the hidden inputs to the other party.


Speaker: Can Deniz ÇAM

Affiliation MSc in Scientific Computing

Advisor: Assoc. Prof.Dr Esma Gaygısız

Co-Advisor: Assoc. Prof. Dr. Hamdullah Yücel

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 11 September 2023; 14:00

Abstract: Social networks are systems where individuals are related to each other to transfer information. While the existence of such networks is easy to observe, how they are formed is a question that is much more complex. In particular, how individual parties decide to make connections without being forced externally, i.e. endogenously, is a topic that needs further exploration. Endogenous network formation, in itself, requires parties to be incentivised to form connections. This study considers a case where a social network is formed by firms that compete in a market and are incentivised to form connections in order to reduce their production costs. Then, the Multiple Connectivity Game, where players may choose to alter all their direct connections as a single move, is introduced. This game models the endogenous network formation as a result of choices made by players to reduce their production costs and thereby increase their profits. The profit attainable by altering all possible connections simultaneously generates a complex optimisation problem. In order to reduce the complexity of such problems, this thesis proposes converting each connection choice to a discrete decision variable and converting the problem to a mixed integer optimisation problem. Moreover, a method to solve this problem using a modification of the branch and bound algorithm is discussed. Applying this method requires treating discrete network properties as continuous values. An interpolation method that is particular to the problem is provided to facilitate this. Finally, the ability of the proposed model to describe network formation is discussed by simulating various systems with different parameters.


Speaker: Sıla ÖZEREN

Affiliation Msc in Cryptography

Advisor: Assoc. Prof. Oğuz Yayla

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 11 September 2023; 11:00

Abstract: As we transition into the quantum computing era, the security of many commonly used cryptographic algorithms faces a significant challenge. Quantum computers can solve complex mathematical problems in ways that compromise the safety of conventional cryptosystems such as RSA, DSA, and elliptic curve cryptosystems. This thesis provides a comprehensive study on the CRYSTALS-Kyber key encapsulation mechanism (KEM), one of the fourth round finalist of NIST's PQC Standardization effort. Beginning with a detailed explanation of the foundational concepts of lattices, we introduce the hard problems inherent in lattice cryptography - Learning with Errors (LWE), Ring-LWE, and Module-LWE. We then delve into the three components of Kyber.CPAPKE and provide the Fujisaki-Okamoto transform version of each algorithm required to achieve the IND-CCA2 security. We conduct a detailed study on masking Kyber to have a secure compression algorithm in CRYSTALSKyber. We propose two methods for masking its compression function. One integrates a lookup- table, and the other utilizes a double-and-check method. Additionally, we present potential compression functions for various prime numbers.


Speaker: Cansu CEYLAN

Affiliation MSc in Financial Mathematics without Thesis

Advisor: Assoc. Prof.Dr Esma Gaygısız

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 11 September 2023; 10:30

Abstract: Gold is a valuable metal that has historically been highly regarded by civilizations. It has been used in jewellery industries, decorative purposes and mainly in finance. Especially in uncertain financial environment investors invest in gold to protect themselves unwanted price movements. Forecasting the rate of return of gold is significant issue in financial environment. This paper aims that forecast and compare the return of gold using different time series model. ARIMA-GARCH, ARIMA-EGARCH, ARIMA-TGARCH, Simple exponential smoothing model, Holt’s linear exponential smoothing model and HoltWinters’ exponential smoothing model has been used and evaluated using by measure of accuracy methods.


Speaker: Murat Burhan İLTER

Affiliation PhD  (or MSc) in Programme Name PhD in Cryptography

Advisor: Assoc. Prof. Dr. Ali Doğanaksoy

Co-Advisor: Prof. Dr. Ali Aydın Selçuk

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 08 September 2023; 15:00

Abstract: The security of block ciphers can be evaluated using cryptanalysis methods. The use of Mixed-Integer Linear Programming (MILP)  has gained prominence due to its effectiveness in analyzing the security aspects of block ciphers. In this thesis, we explore the application of MILP techniques for conducting comprehensive differential and linear cryptanalysis. Our research specifically addresses fundamental challenges in the realm of differential and linear cryptanalysis. In this work, we study the cipher resistance against differential and linear attacks taking into account that ciphers need to be resistant to these attacks.  In this context, aiming to identify the best differential and linear characteristics of a block cipher is a challenging problem. To tackle these challenges, our work introduces innovative MILP modeling methods for equations involving multiple xor operations. These models, denoted as Model 1 and Model 2, offer alternatives with fewer variables and constraints, respectively. Model 1 and Model 2 generally provide shorter solution times compared to the standard xor model. Importantly, these proposed models have broad applicability beyond differential and linear cryptanalysis, enhancing their utility in various cryptanalysis methods. We model well-known ciphers such as KLEIN, PRINCE, FUTURE, and IVLBC with MILP. The resulting models enable us to precisely determine the exact minimum number of active S-boxes, and the best differential and linear characteristics. Applying our developed MILP models provides improvements in the best single-key differential and linear characteristics for the examined ciphers.


Speaker: Fatih AYKURT

Affiliation MSc in Cryptography

Advisor: Assoc. Prof. Dr. Oğuz YAYLA

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 07 September 2023; 10:00

Abstract: Using secure multi-party computing protocols (MPC), a group of participants who distrust one another can securely compute any function of their shared secret in puts. Participants exchange these inputs in a manner similar to secret sharing, where each participant owns a portion of the input but is unable to independently reconstruct the complete information without collaborating with the other participants. This kind of computation is quite powerful and has many uses where data privacy is quite critical such as areas like government, business, and academia. MPC has grown from a subject of theoretical study to a technology being employed in industry, becoming effective enough to be deployed in practice. In this work two versatile MPC frameworks, MP-SPDZ and MPyC profiling analysis are studied. From basic operations to more complex structures, the bottleneck of the algorithms where the time consumption increases drastically are shown. According to detailed experiments and implemented algorithms, found that the MPyC bottleneck’s effect on its performance is outrageous. Besides that MP-SPDZ has an unexpected profiling outcome on the sorting algorithm. Profiling results are also visualized as dot graphs to be able to detect the critical parts easily.


Speaker: Yeşim GİRGİN

Affiliation Msc in Financial Matematics

Advisor: Prof. Dr. A.Sevtap Kestel

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 05 September 2023; 15:30

Abstract: Modeling exchange rate volatility is a major concern for researchers, investors, and policymakers since it has a wide-ranging impact on the country’s economy, including inflation, interest, investment, production, and foreign commerce (Saglam, 2016). Therefore, the primary goal of this research is to model the volatility of the exchange rate. For this purpose, the generalized autoregressive conditional heteroscedastic techniques comprising of symmetrical (GARCH) and asymmetrical (EGARCH, TGARCH, and APARCH) models are used in this study. Furthermore, aside from the studies conducted in the Turkish literature on that matter regarding models’ distribution, various distributions which consist of skew normal, skew student t, and skew GED along with normal, student t, GED distributions are utilized for the error distribution in GARCH models. The data is taken from CBRT’s closing prices in US dollars consisting of the period of June 2001 to June 2023. Convenient models are put forward based on model selection criteria such as Akaike (AIC), Schwarz (SC), and Log-Likelihood. Finally, the result of this paper concluded that ARMA (4, 3)-EGARCH (1, 1) with skew t distribution is the best model among the proposed models according to the selection criteria.


Speaker: Onur Enginar

Affiliation PhD in Department of Financial Mathematics 

Advisor: Prof. Dr. Ömür Uğur

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 05 September 2023; 11:30

Abstract: In this thesis study, we develop a novel deep ensembles based architecture that enables transfer learning to reduce the time requirement of deep ensembles without compromising accuracy of the model. We apply our model to open energy datasets. Moreover, in this thesis, we compare SoTA tabular learning models with deep ensembles and traditional machine learning model and provide a benchmark. We further develope a feature selection algorithm based on deep ensembles model and compare it with linear feature selection models tree based feature selection algorithms.


Speaker: Aytaç KARA

Affiliation PhD in Financial Mathematics

Advisor: Prof. Dr. Ömür Uğur

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 05 September 2023; 09:30

Abstract: Recent events such as the Brexit, the Covid-19 pandemic and the Russia-Ukraine War, have put a lot pressure on the financial world and markets have become more volatile due to the increase in uncertainty of what the future will hold. Exchange rate series are highly volatile, chaotic and noisy in this environment (particularly Turkish Lira). In the thesis study, we focus on forecasting exchange rates and treat neural networks as an alternative approach and investigate whether they can provide superior forecasts. We particularly study on Recurrent Neural Networks (RNNs), a class of artificial neural networks where connections between nodes can create a cycle. We aim to present the comparison results between performances of RNNs and other statistical forecasting methods. We apply Turkish Lira (TRY), American Dollar (USD) and EURO alongside British Pound (GBP) and Japanese Yen (JPY) against each other using their historical data. We use five different models over three cases: Elman RNN, Encoder-Decoder LSTM, basic type of LSTM, VAR model and VECM. The first two cases go through same calendar spread (Feb 2021-Dec 2022) and focus on Turkish Lira and US Dollar, the currency with less volatility, respectively. The third case covers different period (April 2017-February 2021) and focuses on Turkish Lira again. Technical analysis is favored over fundamental analysis and Optuna is chosen as the hyperparameter optimization method. Results show that even if there is no clear winner model here, recurrent neural networks performed better than statistical models at all three cases. Whether we have shorter or longer calendar spread and more or less volatile currencies, recurrent networks are still ahead of statistical models we have used here.


Speaker: Etkin HASGÜL

Affiliation PhD in Financial Mathematics

Advisor: Prof. Dr. A. Sevtap KESTEL

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 04 September 2023; 13:00

Abstract: This thesis presents a innovative hybrid approach combining Time Series, Artificial Neural Network (ANN), and Copula models for calculating the Solvency Capital Requirement (SCR) concerning Non-Life Premium Risk across multiple lines of businesses (LoB). The loss ratio is formulated as Zi=Xi+Yi+εi where Xi represents the Time Series component, and Yi denotes the ANN component. Initially, the loss ratios are subjected to modeling through suitable time series models to capture the linear component of the model. Subsequently, the appropriate autoregressive neural network (NNAR) model is applied to the residuals resulting from the time series modeling for each LoB, representing the non-linear component. Lastly, the residuals of the combined model are modeled using an R-vine structure. By utilizing these models and the copula structure, simulated loss ratios are generated for the selected LoBs, enabling the analysis of Non-Life Premium Risk. The comparison with the Standard SCR model and the proposed model incorporating VaR and TVaR is performed to assess the efficacy of the proposed approach.


Speaker: İbrahim ÖZBEK

Affiliation MSc in Financial Mathematics

Advisor: Assoc. Prof. Dr. Esma Gaygısız

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 04 September 2023; 10:30

Abstract: This thesis analyzes the monetary policy expectations of various market-based instruments. I investigate which financial instrument best estimates monetary policy expectations for different time periods in Turkiye. A new approach is adopted, and forward-term policy rates are obtained from the yield curve factors. The Nelson-Siegel method, widely used in literature, is preferred while fitting the yield curve. The predictive power of treasury, FX swap, and OIS implied yields are analyzed. Empirical findings reveal that instruments' success in estimating the CBRT policy rate has changed over time. OIS yield curve successfully predicts the monetary policy stance after the Turkish Lira O/N Reference Rate (TLREF) market becomes active.


Speaker: Zeynelabidin KARAKAŞ

Affiliation PhD in Cryptography

Advisor: Prof. Dr. Ferruh ÖZBUDAK

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 31 August 2023; 12:00

Abstract: In this thesis, we aim to improve the current bounds for a certain type of code and not only find the number of distinct codes but also characterize them for some parameters. Flag codes have applications in network coding and their algebraic and combinatorial structures have been an active research area in recent years, see, for example, \cite{alonso,sascha}. Characterization of maximal flag codes of a given type and distance over a given ambient space is a very difficult problem. In this paper, we completely solve this problem for small parameters using also MAGMA. In particular, we find new maximal flag codes as well.


Speaker: Kübra KAYTANCI

Affiliation PhD in Cryptography

Advisor: Prof. Dr. Ferruh ÖZBUDAK

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 31 August 2023; 11:00

Abstract: A desired goal in designing good cryptosystems is to construct boolean functions with good cryptographic properties, such as having high nonlinearity, balancedness, high correlation immunity, and high algebraic immunity. In this thesis, we obtain concrete upper bounds on the algebraic immunity of a class of highly nonlinear plateaued functions without linear structures than the one given recently in 2017 by Cusick. Moreover, we extend Cusick's class to a much bigger explicit class, and we show that our class has better algebraic immunity by an explicit example. We also give a new notion of the linear translator, which includes the Frobenius linear translator given in 2018, Cepak, Pasalic, and Muratovic-Ribic as a particular case. We find some applications of our new notion of linear translator to the construction of permutation polynomials. Furthermore, we give explicit classes of permutation polynomials over Fq2 using some properties of Fq and some conditions of 2011, Akbary, Ghioca, and Wang. Additionally, recently Ellingsen et al. introduced a new concept, the c-Difference Distribution Table and the c-differential uniformity, by extending the usual differential notion. The motivation behind this new concept is based on having the ability to resist some known differential attacks, as shown by Borisov et. al. in 2002. In 2022, Hasan et al. gave an upper bound of the c-differential uniformity of the perturbed inverse function H via a trace function Tr(x2/(x+1) ). In their work, they also presented an open question on the exact c-differential uniformity of H . By using a new method based on algebraic curves over finite fields, we solve the open question in the case Tr(c)=1= Tr(1/c ) ∈ F2n/{0,1} completely and we show that the exact c-differential uniformity of H is 8. In the remaining case, we almost completely solve the problem, and show that the c-differential uniformity of H is either 8 or 9.


Speaker: Dilek ÖNER ŞİMŞEK

Affiliation PhD in Cryptography

Advisor: Prof. Dr. Ersan AKYILDIZ

Co-Advisor: Prof. Dr. Murat CENK

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 29 August 2023; 17:00

Abstract: In the cloud, private data is open to security problems. Homomorphic encryption is the solution to solve this privacy problem by allowing computations on encrypted data. Secure matrix multiplication is one of the primary operations in many applications, including statistical analysis. Security of matrix multiplication comes from homomorphic encryption. The secure matrix multiplication algorithm creates packed polynomials, encrypts them, and multiplies homomorphically. It is hard to do homomorphic multiplication for large matrices since the degree of the packed polynomials increases. In this thesis, we propose using the Strassen algorithm after dividing matrices into subblocks and applying a secure matrix multiplication algorithm to multiply two matrices securely. The Strassen algorithm reduces the number of homomorphi encryptions from eight to seven when one-level recursion is used. Moreover, the homomorphic encryptions and multiplications are performed more efficiently because of a decrease in the degree of polynomials. We implement the Strassen algorithm for secure matrix multiplication using Fan and Vercauteren (FV) and Brakerski, Gentry, and Vaikuntanathan (BGV) homomorphic encryption algorithms. The implementation results for dimensions between 8 and 128 with different submatrix sizes demonstrate significant improvements. To illustrate, when the FV homomorphic encryptio is used for 128 x128 matrix with the submatrix size is two, the algorithm is 47% faster than the standard secure block matrix multiplication method. Similarly, when using the BGV algorithm and considering a 128 x 28 matrix with a submatrix size of two, the algorithm is faster than 44% compared to the standard secure block matrix multiplication. So, our implementation highlights the benefits of using the Strassen metho for secure matrix multiplication within the homomorphic encryption framework emphasizing its potential to improve performance, especially for bigger dimensions after dividing matrices into subblocks. Moreover, as an application, we calculate linear regression of encrypted data using the Strassen secure block matrix algorithm and compare results with the standard secure block matrix method. According to the results when the matrix dimension is 128, and the subblock size is two, calculating linear regression with Strassen’s secure block matrix multiplication algorithm is 47% faster than the standard secure block matrix multiplication algorithm.


Speaker: Anıl Burak GÖKÇE

Affiliation MSc in Cryptography

Advisor: Assoc. Prof. Dr. Oğuz YAYLA

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 29 August 2023; 15:30

Abstract: With the advancements in quantum computing, traditional cryptography is considered to have little life in the future. That’s why NIST initiated a Post-quantum cryptography-related project in order to standardize quantum-secure cryptography. With the latest report on this project, the dominating quantum-secure problems appear to stem from lattice structures. Thus, efficient implementation techniques on polynomial multiplications, which is the bottleneck of lattice-based cryptography, emerged as an important topic. In this thesis, we propose a new 5-way split TMVP algorithm and its application to polynomial multiplications with an implementation of the lattice-based algorithm NTRU KEM. The results are promising, showing up to 34%, 35%, and 157% speed-up against Toom4-Karatsuba implementation in key generation, encapsulation, and decapsulation, respectively.


Speaker: İlksen ACUNALP ERLEBLEBİCİ

Affiliation PhD in Cryptography

Advisor: Assoc.Prof.Dr. Oğuz YAYLA

Place/Zoom Link: https://zoom.us/j/3122100970 Password: 0970

Password: 0970

Date/Time: 20 July 2023, 18:00

Abstract: In this thesis we present our work on the Carlet's bivariate APN construction and exponent of APN power functions. We report necessary and sufficient conditions on some families of bivariate APN functions and our observations about representation of the exponent of the APN power functions by following the methods given by Calderini et al in 2021 and Budaghyan et al in 2022 and we give some experimental data on APN power function. Keywords: APN functions, bivariate APN construction, APN power functions,


Speaker: Hatice Kübra Güner

Affiliation PhD in Cryptography

Advisor: Assoc.Prof.Dr. Oğuz YAYLA

Co-Advisor: Ceyda Mangır

              Place/Zoom Link: Hayri Körezlioğlu Seminar Room

Date/Time: : 20.07.2023 – 16:00

Abstract: Protecting secret keys from malicious observers is a major problem for cryptographic algorithms in untrusted environments. White-box cryptography suggests hiding the key in the cipher code with an appropriate method such that extraction of the key becomes impossible in the white-box setting. The key is generally embedded into the confusion layer with suitable methods. One of them is using encoding techniques. Nevertheless, many encoding methods are vulnerable to algebraic attacks and side-channel analysis. Another is the space hardness concept, which creates large lookup tables that cannot be easily extracted from the device. In (M,Z)-space hard algorithms, the secret key is embedded in large tables created as a substitution box with a suitable block cipher. So the key extraction problem in the white-box setting turns into a key recovery problem in the black-box case. One of the main issues in (M,Z)-space hard algorithms is accelerating the run-time of the white-box/black-box implementation. In this study, we aim to use the advantage of the efficiency of lightweight components to speed up the diffusion layer of white-box algorithms without decreasing the security size. Therefore, we compare the linear layer of NIST Lightweight Standardization candidates for efficiency and suitability to white-box settings in existing space hard ciphers. The performance results of the algorithms are compared with WARX and SPNbox-32. According to the results, using the  lightweight components in the diffusion layer accelerates the performance of white-box algorithms by at least eight times. Additionally, we propose an LS-design based white-box algorithm that has better run-rime performance and an LS-design based table creation method to take advantage of the bitslice implementation against side-channel attacks. When we compare run-time performance of our method with SPNbox algorithm, we obtain 28% improvement for white-box implementation and 27% for black-box implementation. At the same time, in the white-box setting, the LS-design based method is also implemented to the 256-bit block size.


Speaker: Fatma Sıla Kaya

Afiliation Master’s (non-thesis)  in Cryptography

Advisor: Assoc.Prof.Dr. Oğuz YAYLA

Place/Zoom Link: Hayri Körezlioğlu Seminar Room

Date/Time: 18.07.2023 / 16:00

Abstract: In this project, single server private information retrieval (PIR) protocols are examined. An extensive discussion is initiated on the fundamental principles and core concepts underpinning single server PIR schemes. Furthermore, particular emphasis is placed on the introduction of computationally secure single server PIR (CPIR) and computationally symmetric single server PIR (CSPIR) constructions. Important concepts aimed at reducing the overall complexity inherent in PIR protocols are presented as well. Finally, to illustrate the latest advancements in single server PIR field, the project ends with a review of the state-of-the-art SimplePIR scheme.


Speaker: Betül Kalaycı

Afiliation: PhD in Financial Mathematics

Advisor: Prof. Dr. Vilda Purutçuoğlu

Co-Advisor: Prof. Dr. Gerhard-Wilhelm Weber

              Place/Zoom Link: Hayri Körezlioğlu Seminar Room

Date/Time: : 22.06.2023 – 11:00

Abstract: Economists have conducted research on several empirical phenomena regarding the behavior of individual investors, such as how their emotions and opinions influence their decisions. All those emotions and opinions are described by the word Sentiment. Apart from sentiment term, there are also term which is called as Consumer Confidence. This term basically explains the peoples’ feelings about economy according to their confidence levels. In finance and economics, stochastic changes might occur according to investors' and consumers' sentiment levels. One of the aim of this thesis to represent the mutual effects between some financial processes and sentiment indexes with machine learning models. Here, we compare the gain in accuracy and computational time with each of the models' strong alternatives applied in the analyses of the financial data.  Hence, the goal of this thesis is to compare the forecasting performance of sentiment index, consumer confidence index and some other macroeconomic data (CPI, unemployment rate and currency rate) by using first for one-stage MARS (Multivariate Adaptive Regression Splines), NN (Neural Network) and RF (Random Forest) models, then for two-stage MARS-NN, MARS-RF, RF-MARS, RF-NN, NN-MARS, and NN- RF hybrid models.  Furthermore, we aim to use volatility models for mainly sentiment index, consumer confidence index and to observe the relationship with some macroeconomic data (CPI, unemployment rate and currency rate) so that we can get better forecasting results from those datasets. On the other hand, these kinds of series are prone to exhibit significant structural breaks. At this point, we introduce the Markov Switching model and since these breaks include different volatility structures, we merge Markov Switching model with GARCH model, which defines Markov Switching GARCH (MS-GARCH) model.


Speaker: Pelin Çiloğlu

Afiliation: Ph.D. in Scientific Computing

Advisor: Assoc. Prof. Dr. Hamdullah Yücel

              Place/Zoom Link: Hayri Körezlioğlu Seminar Room

Date/Time: 14.06.2023 – 10:00

Abstract: :Uncertainty, such as uncertain parameters, arises from many complex physical systems in engineering and science, e.g., fluid dynamics, heat transfer, chemically reacting systems, underwater pollution, radiation transport, and oil field reservoir.   It is well known that these systems can be modeled by partial differential equations (PDEs) with random input data. However, the information available on the input data is very limited, which cause high level of uncertainty in approximating the solution to these problems. Therefore, the idea of uncertainty quantification (UQ) has become a powerful tools to model such physical problem in the last decade.

In this thesis, the aim is the development, analysis, and application of stochastic discontinuous Galerkin method for partial differential equation (PDE)-based models with random coefficients. As a model, we first focus on the single convection diffusion equation containing uncertainty. To identify the random coefficients, we use the well–known technique Karhunen Loève (KL) expansion.  Stochastic Galerkin (SG) approach, turning the original problem containing uncertainties into a large system of deterministic problems, is applied to discretize the stochastic domain, while a discontinuous Galerkin method is preferred  for the spatial discretization due to its better convergence behaviour for convection dominated PDEs. A priori and a posteriori error estimates are derived. SG method generally results in a large coupled system of linear equations, the solution of which is computationally difficult to compute using standard solvers. We provide low-rank iterative solvers for efficient computing of such solutions, which compute low-rank approximations to the solutions of those systems. Moreover, to overcome boundary and/or interior layers, localized regions where the derivative of the solution is large, an efficient adaptive algorithm is presented for the numerical solution of the parametric convection diffusion equations. On the other hand, certain parameters of a model are need to be optimized in order to reach the desired target, for instance, the location where the oil is inserted into the medium, the temperature of a melting/heating process, or the shape of the aircraft wings. Therefore, we extend our findings to optimization problems and consider  optimal control problems governed by convection diffusion equations involving random inputs. By following the similar numerical approaches as done for the single PDE-based model,  we illustrate the efficiency of the proposed approaches.


Speaker: Yasemin Yaman Kanmaz

Afiliation: PhD in Financial Mathematics

Advisor: Prof. Dr. Sevtap Kestel

              Co-Advisor: Prof. Dr. Kasırga Yıldırak

Zoom Link: https://zoom.us/j/95222106345?pwd=akQ1VGdrL3hvdkw1WW94OFh2L08vUT09

Meeting Number: 952 2210 6345

Password: 004882

Date/Time: 02.06.2023 - 09:30

Abstract: In data sets with imbalanced class distribution, traditional machine learning algorithms tend to label most of the minority-class instances as majority-class ones when there is no specific feature distinguishing the minority class. The unequal class distributions result in two types of prediction errors that incur different costs in imbalanced credit data sets. These are monetary losses for the misclassified defaults and opportunity cost of interest income for the misclassified non-defaults. Addressing these issues, this study proposes a novel approach to cost-sensitive learning and imbalanced data classification in credit data sets, using new borrower (instance)-specific cost/risk parameters to solve the two types of asymmetries. The main objective of this study is to create a weight-signaling risk level for each instance by revealing instance-embedded information to strengthen ordinary algorithms with the generated weight and breaking the dominance of the majority class in the loss functions. The default probabilities of credit applicants provide valuable information about their risk level, and thus new instance-specific cost/risk parameters based on their default risk levels are proposed instead of class-specific ratios. Default probabilities are estimated with sampled sub-datasets, and before this step, analyses for the class ratio of sub-datasets are conducted with the Simulated Annealing stochastic process searching the optimal thresholds maximizing Gmean and evaluating R2 for different class ratios. To estimate the default probabilities, non-linear complex models like logistic regressions, deep learning-based Graph Neural Networks, and Graph Attention Networks are employed. Three cost/risk parameters are generated with the target of equalizing the class losses based on their class-based default risk level aggregations. Credit applicants are re-weighted with new cost/risk parameters based on their default risk probabilities, and the new parameters function as boosting the minority class since higher weights are assigned to the minority class instances due to their relatively higher default risks. AdaBoost, XGBoost, and ANN algorithms are then modified to incorporate these new parameters. The empirical analyses are conducted using eight credit data sets. A detailed comparison between existing algorithms and new cost-sensitive algorithms is conducted. The proposed cost-sensitive algorithms result in more equal Sensitivity and Specificity values, and lower financial losses given the default values for given Specificity values. The success of these algorithms is particularly evident in the classification of data sets where the class ratios increase.


Speaker: Cigdem Yerli

Afiliation: PhD in Financial Mathematics

Advisor: Prof. Dr. Sevtap Kestel

              Co-Advisor: Assoc. Dr. Zehra Eksi-Altay

Zoom Link: https://zoom.us/j/95222106345?pwd=akQ1VGdrL3hvdkw1WW94OFh2L08vUT09

Meeting Number: 952 2210 6345

Password: 004882

Date/Time: 16.05.2023 13:30 

Place: Hayri Körezlioğlu Seminar Room, IAM, METU

Abstract: This thesis presents a methodology for modeling the implied liquidity which is introduced through the Conic Finance theory, and considered a proxy for the market liquidity level. We propose a partial information setting in which the dynamics of implied liquidity, representing the noisy information on the unobserved true market liquidity, follow a continuous-time Markov-chain modulated exponential Ornstein-Uhlenbeck process. Model inference requires the filtering of the unobserved states of the true market liquidity, as well as the estimation of the unknown model parameters. We address the inference problem by the EM algorithm. The expectation step of the algorithm requires the derivation of finite dimensional filters for the quantities present in the likelihood function. To this end, we first review the existing EM algorithm for the OU process and provide detailed proofs. The application of the algorithm in practice needs discretizing the resulting filters. In order to avoid numerical issues and make the algorithm to function, we introduce filters that have a continuous dependence on the observations. The corresponding filters are known as robust filters. Instead of directly discretizing continuous time filters, we discretize the robust filters that help us to work under the discrete time setting and also enable us to obtain the variance estimate of the model within the EM algorithm. We evaluate the performance of the algorithm and compare it to existing alternatives in the literature using an extensive simulation study. The performance evaluation is based on the sensitivity to changes in step size, drift, and volatility parameters. This step is crucial for refining the methods and establishing a connection between theory and practice. Once the algorithm is tested with simulated data, we apply the proposed model to real world data. The data set comprise of implied market liquidity series retrieved from the S&P 500 option data covering the period from January 2002 to August 2022. Our application results show that three liquidity regimes can describe the market liquidity level: high, intermediate and low. The estimation results confirm the effect of the overall economic environment on the market liquidity.


Speaker: Ceylin Doğan

Afiliation: MSc in Cryptography

Advisor: Assoc. Prof. Dr. Oğuz Yayla

Zoom Link: https://zoom.us/j/3122100970?pwd=eXBKL2l2VVlpcFpvOXhWcEk5bC9qQT09

Meeting Number: 3122100970

Password: 0970

Date/Time: 28.04.2023 16:00 (the thesis defense will take place in class; is also available as virtual on zoom)

Place: Hayri Körezlioğlu Seminar Room, IAM, METU

Abstract: Timed-release cryptography is an innovative approach to sending information that is designed to be received at a specific time in the future. The thesis focuses on the evolution of timed-release cryptography, which was initially proposed by May in 1993 as a means of sending encrypted messages into the future. This concept led to the development of time-lock puzzles by Rivest, Shamir, and Wagner in 1996, which enabled the generation of puzzles with hidden solutions that became visible after a specific time had elapsed. This thesis provides a comprehensive survey of the existing literature on timed cryptography, including time-lock puzzles, timed commitment schemes, and timed signature schemes, to provide a historical background of timed cryptography. In addition, this study analyzes the efficiency and security levels of various cryptographic techniques, identifies areas for future research and development, and highlights the potential applications of timed cryptography in real-life scenarios such as contract signing and payment channel protocols.


Speaker: Ceyda Çaylak

Afiliation: MSc in Financial Mathematics

Advisor: Assoc. Prof. Dr. Seza Danışoğlu

Zoom Link: https://zoom.us/j/99516157521?pwd=Znl5VVRoYUtiMW93Y3JMb2ZHUlFiUT09

Meeting Number: 995 1615 7521

Password: 017035

Date/Time: 28.04.2023 11:15 

Place: Hayri Körezlioğlu Seminar Room, IAM, METU

Abstract: : In finance theory, investors are assumed to be risk averse, behave rationally, and optimize their returns. However, anomalies, such as excess trading volume or excess volatility, are observed in the market and these anomalies need to be explained in light of the rationality assumption. Behavioral finance theories suggest that overconfidence is a notable bias that may explain some of these anomalies. Overconfident investors are more inclined to attribute their success to their abilities and knowledge rather than luck or the announcements in the market. In this study, stock preferences of individual and institutional investors and the outcome of their investment choices are examined during bull and bear periods by using panel and regression analyses. Findings of the study suggest that individual ownership is higher for stocks with higher volatility and book-to-market values during bull periods, implying that these investors are more likely to hold the stocks that they think are undervalued but might be more valuable in the future. Besides, this relationship is even stronger during bear periods. Results show that the institutional investors prefer stocks with low volatility and book-to-market during both bull and bear periods. The study leads to the conclusion that individual and institutional investors behave differently under similar market environments and suffer from psychological biases in the Turkish stock market. For future studies, a bull and bear determination model that focuses on the substantial rises or falls in the stock market over an extended period can be considered.


Speaker: İlayda Tosun

Afiliation: MSc in Financial Mathematics

Advisor: Prof. Dr. Ceylan Yozgatlıgil

Date/Time: 27.04.2023 10:00

Place: Hayri Körezlioğlu Seminar Room, IAM, METU

Abstract: Seasonal adjustment is a statistical technique to remove the effects of regular, periodic fluctuations from time series. Seasonal adjustment is essential for accurately interpreting and analyzing time series, particularly in macroeconomics, where understanding long-term trends and patterns is crucial for making informed decisions. In recent years, there has been a significant increase in the amount of high-frequency time series, which refers to data collected at very short intervals, due to technological developments and increased awareness of the value of data. However, high-frequency time series have unique challenges for seasonal adjustment. These data often have more noise and higher volatility levels. This makes it harder to identify and remove seasonal effects accurately. In this thesis, complex seasonal time series will be examined and will try to adjust them seasonally by comparing current methods.


Speaker: Hüseyin Avni Yaşar

Afiliation: MSc in Scientific Computing

Advisor: Assoc. Prof. Dr. Ercan Gürses

Co-Advisor: Assoc. Prof. Dr. Hamdullah Yücel

Date/Time: April 14, 2023 at 15:30

Place: Hayri Körezlioğlu Seminar Room, IAM, METU

Abstract: This thesis presents a novel approach for predicting the buckling load of stiffened panels using multi-fidelity modeling based on the quadratic neural networks (QNNs) with adaptive activation functions. The effectiveness of the proposed approach is demonstrated through a series of simulations on a range of stiffened panel configurations, and the results are compared to those obtained from traditional multi-fidelity modeling methods in terms of accuracy and computational efficiency. Numerical experiments demonstrate that the model can accurately and efficiently predict the buckling load of stiffened panels, while significantly reducing the computational cost of evaluating the surrogate model. This approach can significantly improve the design and optimization of aerospace structures by easily and quickly exploring various design configurations and finding stable and efficient configurations. Overall, this study highlights the potential of multi-fidelity modeling for predicting the buckling load of aerospace structures, and the effectiveness of using QNNs with the adaptive activation functions.


Speaker: Mert Malkoç

Afiliation: MSc in Financial Mathematics

Advisor: Assoc. Prof. Dr. İlkay Şendeniz Yüncü

Date/Time: April 10, 2023 at 11:00

Place: Hayri Körezlioğlu Seminar Room, IAM, METU

Abstract: In this thesis,  we aim to see the changes in the various balance sheet items of the Turkish banking sector and the exchange rate between January 2008 and December 2021. The analysis of balance sheet items is conducted in the thesis including the Granger Causality method. Other than balance sheet items, the ratio of foreign loans to total assets, and the ratio of Turkish Lira denominated deposits to total credits are also included in the thesis to assess the changes in the balance sheet items of the Turkish banking sector. Our results show that in the face of increasing exchange rates, the banks try to limit their foreign borrowings and started to increase their deposits to fund the credit channels.


Speaker: Derya Genç

Afiliation: MSc in Financial Mathematics without Thesis

Advisor: Prof. Dr. Ömür Uğur

Zoom Link: Place/Zoom Link: https://zoom.us/j/99052772621?pwd=Snkza1lMWDNpb1ppV2xhQ01nNlh0dz09

Meeting number: 990 5277 2621

Password: 627494

Date/Time: February 03, 2023 at 10:30

Place: Hayri Körezlioğlu Seminar Room, IAM, METU

Abstract: Control are the measures taken by the management and other relevant units to manage risks and increase the likelihood of achieving the determined goals and objectives. These control processes are generally determined by certain frameworks and one of the most valid frameworks is COSO. Along with the pandemic, the risks that banks may face have partially changed. Therefore, control processes should change against these risks. In this study, it has been shown which COSO component these risks belong to and that there should be a dynamic control process with changing risks.


Speaker: Gökberk Yıldırım

Afiliation: MSc in Cryptography

Advisor: Prof. Dr. Ferruh Özbudak

Date/Time: January 27, 2023 at 16:00

Place: Hayri Körezlioğlu Seminar Room

Abstract: This paper represents a routing algorithm that uses routing algorithms techniques to find the shortest routing path. The basic idea behind pathfinding is searching a graph, starting at one point, and exploring adjacent nodes from there until the destination node is reached. Generally, the goal is of course to obtain the shortest route to the destination. This paper searches routing algorithms like the well-known Dijkstra’s Single-source shortest path algorithm. The main objective of this paper is to reduce path-request blocking and increase overall utilization. In routing algorithms, they are reviewed and formulised with uncertainty which comes from weights on edges according to the actual situation on the road such as weather conditions, and road capacity at the specified time. The two key issues that need to be addressed in SPP (Shortest Path Algorithm) are to determine the addition of two edges and to compare the distance between two different paths with their edge lengths.


Speaker: Nika Rasoolzadeh

Afiliation: MSc in Scientific Computing

Advisor: Assoc. Prof. Dr. Yeşim Serinağaoğlu Doğrusöz

Date/Time: January 27, 2023 at 14:00

Place: Hayri Körezlioğlu Seminar Room

Abstract: The inverse problem of electrocardiography refers to the determination of the electrical activity of the heart from the body surface potential measurements (BSPM). Knowledge of the electrical activity state of the heart can provide valuable insights for the diagnosis of cardiac disorders and aid in the facilitation of the development of appropriate treatments. Consequently, efficient resolution of this problem has the potential to be of significant benefit to clinical practices, making it imperative to continue the research in this field. This problem is considered ill-posed due to the presence of attenuation and spatial averaging effects within the thorax between the cardiac sources and the body surface. This characteristic can result in highly inaccurate or oscillatory solutions, particularly in the presence of measurement noise or an imprecise mathematical model. Despite significant efforts to address the challenges and limitations, an optimal method for the analysis of clinical data in this problem has yet to be developed. A relatively limited number of studies have been conducted on the application of inverse problem of electrocardiography for the localization of spontaneous Premature Ventricular Contraction (PVC). PVCs refer to abnormal cardiac contractions that originate from one of the ventricles. This study focuses on using Tikhonov regularization method for reconstruction of unknown heart potentials with respect to various torso volume and cardiac source models derived from BSPMs of patients who have been indicated for ablation. by estimating Activation Time (AT) sequences, the PVC origin locations are explored for a number of patients. The findings indicate that the implementation of the proposed methods could provide physicians with a more accurate localization of spontaneous PVCs.


Speaker: İlayda Tosun

Afiliation: MSc in Financial Mathematics

Advisor: Prof. Dr. Ceylan YOZGATLIGİL

Date/Time: January 27, 2023 at 11:30

Place: Hayri Körezlioğlu Seminar Room

Abstract: Seasonal adjustment is a statistical technique to remove the effects of regular, periodic fluctuations from time series. Seasonal adjustment is essential for accurately interpreting and analyzing time series, particularly in macroeconomics, where understanding long-term trends and patterns is crucial for making informed decisions. In recent years, there has been a significant increase in the amount of high-frequency time series, which refers to data collected at very short intervals, due to technological developments and increased awareness of the value of data. However, high-frequency time series have unique challenges for seasonal adjustment. These data often have more noise and higher volatility levels. This makes it harder to identify and remove seasonal effects accurately. In this thesis, complex seasonal time series will be examined and will try to adjust them seasonally by comparing current methods.


Speaker: Ceylin Doğan

Afiliation: MSc in Cryptography

Advisor: Assoc. Prof. Dr. Oğuz Yayla

Place/Zoom Link: https://zoom.us/j/3122100970?pwd=eXBKL2l2VVlpcFpvOXhWcEk5bC9qQT09

Meeting number: 3122100970

Password: 0970

Date/Time: January 27, 2023 at 10:00

Abstract: Timed-release cryptography is an innovative approach to sending information that is designed to be received at a specific time in the future.  In this thesis, we focus on studying and analyzing existing timed signatures in the field of cryptography. We examine various types of timed signatures and the unique features and characteristics of each one. To this end, we first explore the prominent primitives of timed signatures, namely time-lock puzzles, where a sender publishes a puzzle, the solution of which is the message to be sent, but it remains hidden until a sufficient amount of time has passed to solve the puzzle. Additionally, we also examine timed commitment schemes, which allow the receiver to recover the committed value without the sender's assistance, even in the event of a forced opening phase. We studied classical methods of time-lock puzzles such as repeated squaring and randomized encodings, and also introduce the concept of homomorphic time lock puzzles. To conclude, we examine an efficient verifiable timed linkable ring signature application on the blockchain Monero, a privacy-preserving cryptocurrency, which is a prime example of how timed-signatures can be used in practice.


Speaker: İsmail Onur Kızıloğlu

Afiliation: MSc in Actuarial Sciences

Advisor: Prof. Dr. Ayşe Sevtap KESTEL

Co-Advisor: Dr. Bükre YILDIRIM KÜLEKCİ

Place/Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: January 27, 2023 at 10:00

Abstract: Pricing is one of the crucial aspects of the insurance industry. There are compulsory regulations applied to MTPL insurance in Turkey. After the cap premium application that came with these regulations, the significance of correct pricing for the profitability of the companies has increased even more. In this thesis, premiums are estimated quarterly using Bühlmann and Bühlmann-Straub credibility methods for the cities of Turkey. The cities are divided into groups while finding the credibility results. First, the cities are sorted according to the total number of claims and divided into at least 3 groups and at most 10 groups. The reasons for sorting by the number of claims is that the exposure measure is used for Bühlmann-Straub credibility, and cities that show similarity according to claim count are grouped together. Secondly, credibility results are calculated for cities in which grouping is made by geographical regions. After that, cities are divided into two, three, six, and nine groups by applying K-means clustering and credibility results are computed for cities. As a result of grouping and calculations, results and groups are compared to find best fit group and the best method for Turkey MTPL dataset.


Speaker: Duygu Özden

Afiliation: PhD in Cryptography

Advisor: Doc. Dr. Oguz Yayla

Place/Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: January 27, 2023 at 09:00

Abstract: Digital signatures are a fundamental algorithms in blockchain mechanism, used mainly to authenticate transactions. When users submit transactions, they must prove that they are authorized to spend those funds, while preventing other users from spending them. While there are commonly used signatures schemes such as ECDSA, Schnorr, BLS etc. , verifiability and accountability of the signature schemes are still hot-topic to evaluate and investigate. Timed signatures are good alternatives to create such system as they have efficient versions which are verifiable and applicable to different signature schemes. On the other hand, multisignature schemes are the ones we are interested as accountability can be satisfied by using different types of protocols. In this study, we are planning to construct a timed version of multi signature schemes with an efficient way and applicable to blockchain infrastructure.


Speaker: : İlksen Acunalp Erleblebici

Afiliation: PhD in Cryptography

Place / Zoom :  Hayri Körezlioğlu Seminar Hall, IAM, METU / IAM COLLOQUIUM

Place/Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: January 26, 2023 at 10:00

Abstract: The Carlet’s bivariate APN construction is studied in this thesis. We present necessary and sufficient conditions on some families of bivariate and biprojective APN functions. The methods given by Calderini et al in 2021 and by Gölo ̆glu in 2022 are followed. Then, new conditions are proved and a new family of APN functions is constructed.


Speaker: Hatice Kübra Güner

Afiliation: PhD in Cryptography

Advisor: Oğuz Yayla

Co-Advisor: Ceyda Mangır

Place/Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: January 26, 2023 at 09:00

Abstract: Protecting secret keys from malicious observers is a major problem for cryptographic algorithms in untrusted environments. White-box cryptography suggests hiding the key in the cipher code with an appropriate method such that extraction of the key becomes impossible in the white-box settings. The key is generally embedded into the confusion layer with suitable methods. One of them is using encoding techniques. Nevertheless, many encoding methods are vulnerable to algebraic attacks and side-channel analysis. Another is the space hardness concept, which creates large lookup tables that cannot be easily extracted from the device. In (M,Z)-space hard algorithms, the secret key is embedded in large tables created as a substitution box with a suitable block cipher. So the key extraction problem in the white-box settings turns into a key recovery problem in the black-box case. One of the main issues in (M,Z)-space hard algorithms is accelerating the running time of the black-box/white-box implementation. In this study, we aim to use the advantage of the efficiency of lightweight components to speed up the diffusion layer of white-box algorithms without decreasing the security size. Therefore, we compare the linear layer of NIST Lightweight Standardization candidates for efficiency and suitability to white-box settings in existing space hard ciphers. The performance results of the algorithms are compared with WARX and SPNbox-32. According to the results, using the  lightweight components in the diffusion layer accelerates 


Speaker: Tugay Dağlı

Afiliation: MSc in Scientific Computing

Advisor: Assoc. Prof. Dr. Önder Türk

Date/Time: January 25, 2023 at 14:00

Place: Hayri Körezlioğlu Seminar Room

Abstract: In this thesis, edge-based finite element method (FEM) approximations of Maxwell's equations describing the relationship between the space variables and sources along the electromagnetic field are considered. In particular, the lowest-order Nédélec basis functions are implemented to construct the FEM model of the Maxwell source problem, Maxwell eigenvalue problem (EVP), and electromagnetic wave propagation problem. A computational model is constructed to conduct all these problems in the same framework of an approximation formalism. 

The convergence properties of the Maxwell EVP formulation are analyzed by applying the spectral theory with those of the associated boundary value source problem. Therefore, the analysis for both of these problems are given together with the corresponding numerical results validating the theoretical features. Moreover, a comparison between the two convergent FEM approximations of Maxwell EVP that utilizes the lowest-order Nédélec and the lowest-order Lagrange basis functions is performed. This comparison is done by using a special triangulation, namely, a Powell-Sabin type, of the domain that contains a strong singularity.

The electromagnetic wave propagation problem is also considered using two different approaches, namely, a direct time-domain approximation method and a modal analysis technique. For both approaches, a FEM model of the wave propagation problem is obtained by discretizing the spatial domain using the lowest-order Nédélec basis functions. The time domain approximation of this problem is obtained by employing a finite difference (FD) scheme to approximate the second-order temporal derivative in the obtained FEM model. On the other hand, the frequency domain approximation is acquired by truncating the modal expansion solution. Here, it is set forth that the solution to the electromagnetic wave propagation problem can be represented by an expansion of the approximate eigenmodes that are obtained from the associated Maxwell EVP. As a consequence, it is shown that by exploiting the numerical test cases of homogeneous and inhomogeneous wave propagation problems that both methodologies lead to accurate approximations that agree well with each other.


Speaker: Serhat Sağdıçoğlu

Afiliation: PhD in Cryptography

Advisor: Assoc. Prof. Ali Doğanaksoy

Place/Zoom: Hayri Körezlioğlu Seminar Hall, IAM, METU /  IAM COLLOQUIUM

Zoom Link: https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: January 25, 2023 at 10:30

Abstract: Random numbers are essential for security. However, randomness requirements of security applications are quite different from what is generally known for statistics or simulation purposes. After a long standardization study resulting in a few drafts and several comments from the Public, NIST adopted a diverse battery of entropy estimators (tests) in SP 800-90B to validate the quality of entropy sources used for randomness. My thesis focuses on entropy estimation as per NIST SP 800-90B (Recommendation for the Entropy Sources Used for Random Bit Generation).


Speaker: Hüseyin Avni Yaşar

Afiliation: MSc in Scientific Computing

Advisor: Doç. Dr. Ercan Gürses

Co-Advisor:Doç. Dr. Hamdullah Yücel

Date/Time: January 24, 2023 at 16:30

Place: Hayri Körezlioğlu Seminar Room

Abstract: This thesis presents a novel approach for predicting the buckling load of stiffened panels using multi-fidelity modeling based on the quadratic neural networks (QNNs) with adaptive activation functions. The effectiveness of the proposed approach is demonstrated through a series of simulations on a range of stiffened panel configurations, and the results are compared to those obtained from traditional multi-fidelity modeling methods in terms of accuracy and computational efficiency. Numerical experiments demonstrate that the model can accurately and efficiently predict the buckling load of stiffened panels, while significantly reducing the computational cost of evaluating the surrogate model. This approach can significantly improve the design and optimization of aerospace structures by easily and quickly exploring various design configurations and finding stable and efficient configurations. Overall, this study highlights the potential of multi-fidelity modeling for predicting the buckling load of aerospace structures, and the effectiveness of using QNNs with the adaptive activation functions.


Speaker: Doğan Aktaş

Afiliation: MSc in Financial Mathematics without Thesis

Advisor: Assoc. Prof. Dr. Seza Danışoğlu

Place/Zoom Link: https://us05web.zoom.us/j/89233426101?pwd=YW41YnUwb2FUODlva0ZGV2NGT05BUT09

Meeting number:  892 3342 6101

Password: Rpqf1P

Date/Time: January 24, 2023 at 09:30

Abstract:  Financial markets are very important because of the facility provides in trade. Hence, performance of a financial market is important. Liquidity is a crucial indicator to evaluate a financial market’s performance. In this paper, liquidity performance measures for bond markets are examined with real-life examples.

Speaker: Ahmet Umur Özsoy

Afiliation: : PhD in Financial Mathematics

Advisor: Prof. Dr. Ömür Uğur

Join Zoom Meeting: https://zoom.us/j/98910905073?pwd=T3czSlpwTVJlazZpekF5bHkxaWRpZz09

Meeting ID: 989 1090 5073

Passcode: 712502

Date/Time: January 23, 2023 at 10:30

Place: Hayri Körezlioğlu Seminar Room

Abstract: In our study, we make use of reinforcement learning by considering European options written on the exchange rates. For this purpose, we represent and later reformulate an option pricing model, the QLBS model, taking its roots from reinforcement learning. We first devise the QLBS model to be presented in exchange rates. We provide numerical illustrations as well as further insights in regard to exchange rate markets. In second part, we offer several extensions with one developed theoretically yet left as a further study. We suggest different state construction and later on test our extension in comparison. In a later part of this, we also employ jumps that could occur and observe how the QLBS model reacts. In the third part, we reformulate the QLBS model under market impacts by introducing a large agent whose transactions leave a permanent impact in the foreign exchange rate markets. We offer a completely different approach with the aims of enrichment and contribution to the machine learning in option pricing under market impacts. Through all parts, we obtain convergences to the benchmark prices, and discuss that our reformulations and extensions based on the QLBS model could be an alternative to the traditional option pricing models.


      Speaker:M.Hakan Solmaz

Afiliation: PhD  in Cryptography

Advisor: Prof. Dr. Ersan AKYILDIZ

Zoom Link:https://zoom.us/j/92462211844?pwd=OWJwWWlscmVZNWt3MlFId3p1NTVZQT09

Date/Time: 11.01.2023 – 14:00

Place: Hayri Körezlioğlu Seminar Hall, IAM, METU

Abstract: In most of the electronic communication devices that surround us, advanced cryptographic algorithm needs are implemented on special hardware. These specializedhardware are divided into application-specific integrated circuit (ASIC) and field programmable gate arrays (FPGA). In this thesis, we present a flexible and compact elliptic curve coprocessor for different FPGA platforms. The coprocessor is equipped with a micro-instruction controller to manage point multiplication equations in prime fields. Modular multiplication is the fundamental operation that determines the overall performance of such architectures. Our solution includes a Montgomery modular multiplier that is independent of FPGA vendor. The design has the flexibility to perform almost any practical elliptic curve equation. FPGAs are also used for crypanalysis purposes. In the second half of the thesis, we worked on the implementation of prime factorization problem on FPGA. In this section, we worked on the implementation of the Elliptic Curve Method(ECM) on FPGA. ECM is a factorization method that can be implemented in parallel. Solutions for the management of elliptic curve cores implemented in parallel are presented and their performances are examined.


      Speaker:Mert Malkoç

Afiliation: : Financial Mathematics

Advisor: İlkay Şendeniz Yüncü

Co-Advisor:

Zoom Link:

Meeting ID: 

Password: 

Date/Time: 11.01.2023 – 13:00

Place: Hayri Körezlioğlu Seminar Room

Abstract: In this paper, we aim to see the effects of the foreign exchange (FX) debt level of Turkish banks on the riskiness of the banking sector between January 2008 and December 2021. We use a volatility measure, which consists of two parts: the ratio of foreign loans to total assets, and the ratio of Turkish Lira denominated deposits to total credits. Our results show that an increase in the ratio of foreign loans to total assets triggers an increase in the riskiness of the banking sector. However, the ratio of Turkish Lira denominated deposits to total credits does not have a significant effect on the volatility measure


      Speaker:Yasemin Öztürk

Afiliation: : Msc in Financial Mathematics

Advisor: 

Co-Advisor:

Zoom Link:

Meeting ID: 

Password: 

Date/Time: 06.01.2023 – 09:30

Place: Hayri Körezlioğlu Seminar Room

Abstract: In 2008, the whitepaper titled "Bitcoin: A Peer-to-Peer Electronic Cash System" introduced an original design of a decentralized digital currency. The innovations, described by the pseudonym Satoshi Nakamoto, have led to the evolution of cryptocurrencies and have attracted significant attention from the media and investors. The price of Bitcoin has captured the interest, and the dynamics of Bitcoin have became a popular topic in academia. This thesis examines the dynamics of Bitcoin prices with several network metrics and financial assets. Two periods, chosen by Prophet's change point detection, are studied to capture the relationships in different characteristics. Cointegration relationships are analyzed, and according to the results, VAR and VEC models are estimated for granger causality. The cointegration relationship between BTC and hashrate revealed a unidirectional granger-causality from BTC price to hashrate. Finally, ARIMA, exponential smoothing, and Prophet are used to forecast Bitcoin prices, and the results are compared to find the most accurate forecast model by some accuracy metrics.


      Speaker:İsmail DAŞ

Afiliation: : Msc in Financial Mathematics

Advisor: Prof. Dr. A. Sevtap Selçuk-Kestel

Co-Advisor: Dr. Bilgi YILMAZ

Zoom Link:

Meeting ID: 

Password: 

Date/Time: 02.12.2022 – 12:30

Place: Hayri Körezlioğlu Seminar Room

Abstract: In 2008, the whitepaper titled "Bitcoin: A Peer-to-Peer Electronic Cash System" introduced an original design of a decentralized digital currency. The innovations, described by the pseudonym Satoshi Nakamoto, have led to the evolution of cryptocurrencies and have attracted significant attention from the media and investors. The price of Bitcoin has captured the interest, and the dynamics of Bitcoin have became a popular topic in academia. This thesis examines the dynamics of Bitcoin prices with several network metrics and financial assets. Two periods, chosen by Prophet's change point detection, are studied to capture the relationships in different characteristics. Cointegration relationships are analyzed, and according to the results, VAR and VEC models are estimated for granger causality. The cointegration relationship between BTC and hashrate revealed a unidirectional granger-causality from BTC price to hashrate. Finally, ARIMA, exponential smoothing, and Prophet are used to forecast Bitcoin prices, and the results are compared to find the most accurate forecast model by some accuracy metrics.


      Speaker:Murat Özenç Mert

Afiliation: : PhD in Financial Mathematics

Advisor: Prof. Dr. A. Sevtap Selçuk-Kestel

Zoom Link: https://zoom.us/j/96333373848?pwd=WnNuL2E4bWo4UmR0N2NsbmpJZ2pGQT09

Meeting ID: 963 3337 3848

Password: 878485

Date/Time: 02.12.2022 – 14:00

Place: Hayri Körezlioğlu Seminar Room

Abstract: Insurance markets plays an essential role in the economy of world and its structure requires reinsurance policies due to the growth in populations, extreme (catastrophic) events, political and economical perspectives. In this thesis, stop-loss contracts, one of the reinsurance policy types, are covered for two different contract types: (i) contracts with retention and (ii) contracts with both retention and cap (maximum). This thesis covers two different methodologies, distributional and stochastic behaviors of the claim amounts for the analysis of loss modeling, the costs of insurer and reinsurer, exposure curves to obtain fair premium share. Unlike most studies on the reinsurance policies, the thesis makes an emphasis of time-dependent and time influenced structure of claims and gives comprehensive derivations to model claims amounts and to examine the costs of parties and the exposure curves. In the distributional approach, heavy tailed distributions, specifically, Pareto, Gamma, and Inverse Gamma, are used and the costs of parties and the exposure curves are derived analytically under the selected distributions. Using Monte Carlo simulations and considering the joint analysis of parties’ loss ratios, the optimal retention and maximum levels are found and compared with the values minimizing the risks of parties under VaR and CVaR risk measures. In the stochastic modeling approach, in order to express both random and time-dependent mechanism of the claim amounts, Geometric Brownian Motion with time-varying parameters is used and the costs of parties and the exposure curves are derived analytically since the time elapses during contract period brings dissimilarties on the claim behavior so does on the cost, premium share. Furhtermore, Pareto- Beta stochastic jump diffusion (PBJD) model and its theory behind is implemented for capturing possible extreme losses. This combines collects the derivations for the costs and the exposure curves under PBJD. The emphasis on the applications of reallife data, specifically Turkey’s compulsory traffic insruance claims, is made for the stochastic approaches. The results for the expected costs, the exposure curves are presented. In order to obtain the forecasts values of the loss amounts, the expected costs, and the exposure curves, the time varying parameters are taken as time series and ARIMA family models and cubic spline extrapolation are applied on these series in order to keep the structure of stochastic models.


      Speaker:Mutlu Çelik

Afiliation: Scientific Computing

Advisor: Prof. Dr. Ömür Uğur

Co-Advisor: Prof. Dr. Sinan Eyi

Zoom Link:

Meeting ID: 

Password: 

Date/Time: 01.12.2022 / 16:00

Place: Hayri Körezlioğlu Seminar Room / Room No: 212, IAM

Abstract: : In this thesis, open source softwares SU2 Multiphysics and The Porous material Analysis Toolbox (PATO) based on OpenFOAM are used to conduct loosely coupled analysis of hypersonic non-equilibrium flow and thermochemical ablation. Both solvers have become prominent open source softwares with numerous validation and verification cases. NEMO, the non-equilibrium modeling solver of SU2 is used to model chemically reactive and non-equilibrium flows by integrating thermochemistry library of Mutation++. SU2-NEMO solves Navier Stokes equations with thermochemical non-equilibrium effects by using finite volume method. As a material response code, ablation solver PATO is used to calculate the surface temperature of solid materials and surface recession due to ablation. PATO, as a fully portable OpenFOAM library, discretizes conservation equations of total energy, gas momentum, gas mass, solid mass and gas species equations by finite volume method. The outputs of surface temperature and recession are used as inputs of SU2-NEMO for CFD analysis. SU2-NEMO then calculates the surface heat flux and it is used as input of PATO. At the end, the effect of surface recession to surface heat flux distribution of blunt nose geometry is investigated.


       Speaker:Sarp Tuğberk Topallar

       Affiliation: METU

              Advisor: Prof. Dr. Ceylan Yozgatlıgil

              Co-Advisor

Zoom Link: 

Meeting ID: 

Password: 

Date/Time: 02.09.2022 – 14:00

Place: Hayri Körezlioğlu Seminar Room

Abstract: Time series forecasting can be summarized as predicting the future values of a sequence indexed by timestamps based on the past records of that sequence. Optimal or near-optimal resource allocation requires accurate predictions into the future. The proposed solution is to model many similar series with a single common model in order to alleviate the need for synthetic data for augmentation thus obtaining a sample efficient RNN model. In order to observe the effects of information sharing among many series on the model accuracy, this study includes inspection of classical methods including Autoregressive Integrated Moving Average (ARIMA), Exponential Smoothing (ETS) and Seasonal-Trend decomposition using LOESS (STL) as well as more contemporary machine learning methods such as DeepAR which is a Recurrent Neural Network (RNN) comprised of Long Short Term Memory (LSTM) cells. The training data set is comprised of more than one hundred and twenty time series of daily demand records spanning two years for physical stores throughout Türkiye. The provided error metrics for accuracy measurement are MAE (Mean Absolute Error) in order to show the actual value of the forecasted demand and MAPE (Mean Absolute Percentage Error) in order to compare the results with other models independent of the scale. The aim of the model is to provide accurate and robust probabilistic forecasts. The probabilistic forecasts are obtained by training the model in order to learn a probability distribution and producing the point forecasts from sampling the learned probabilistic function. The probabilistic forecasts of different quantiles provide practicality such as forecasting in different quantiles per different series in the same data set in order to provide further fine tuning for many series. Additionally the RNN model does not have the modeling assumptions of the classical models. These assumptions would have required inspection and confirmation per series basis which would greatly decrease the practical applicability. Robustness of the model is inspected in two aspects, the first one is to provide robustness in terms of homogeneity of errors for each of the series in the training data and some true out-of-sample predictions in order to observe the performance of the cold start forecasts generated by the model. The second robustness criterion is the non-degrading performance of the model when the forecast horizon extends further into the future thus eliminating the need for frequent retraining. The study covers the data set analysis, data preprocessing, model training, and hyperparameter tuning, as well as the discussion of the results and the probable future work.


       Speaker: Mutlu ÇELİK

       Affiliation: Scientific Computing

              Advisor: Prof. Dr. Ömür Uğur

              Co-Advisor: Prof. Dr. Sinan Eyi

Zoom Link: 

Meeting ID: 

Password: 

Date/Time: 02.09.2022 / 11:30

Place: Hayri Körezlioğlu Seminar Room / Room No: 212, IAM

Abstract: In this thesis, open source softwares SU2 Multiphysics and The Porous material Analysis Toolbox (PATO) based on OpenFOAM are used to conduct loosely coupled analysis of hypersonic non-equilibrium flow and thermochemical ablation. Both solvers have become prominent open source softwares with numerous validation and verification cases.

NEMO, the non-equilibrium modeling solver of SU2 is used to model chemically reactive and non-equilibrium flows by integrating thermochemistry library of Mutation++. SU2-NEMO solves Navier Stokes equations with thermochemical non-equilibrium effects by using finite volume method.

As a material response code, ablation solver PATO is used to calculate the surface temperature of solid materials and surface recession due to ablation. PATO, as a fully portable OpenFOAM library, discretizes conservation equations of total energy, gas momentum, gas mass, solid mass and gas species equations by finite volume method. The outputs of surface temperature and recession are used as inputs of SU2-NEMO for CFD analysis. SU2-NEMO then calculates the surface heat flux and it is used as input of PATO. At the end, the effect of surface recession to surface heat flux distribution of blunt nose geometry is investigated.


       Speaker: Mehmet Emin Gülşen

Affiliation: MSc in Cryptography

Advisor:  Assoc. Prof. Oğuz Yayla

Zoom Link: : https://zoom.us/j/3122100970?pwd=eXBKL2l2VVlpcFpvOXhWcEk5bC9qQT09#success

Meeting ID: 312 2100 9702

Password: 

Date/Time: Thursday, September 01, 2022 ; 12:00

Place: Hayri Körezlioğlu Seminar Room, IAM

Abstract: Vehicle Routing Problem (VRP) is a classical combinatorial optimization and integer programming problem. The goal of VRP is to find the optimal set of routes to given set of destination points with a fleet of vehicles. In this thesis, we have focused on the a variant of VRP which is Capacitated Vehicle Routing Problem (CVRP) and present two different combination of heuristic algorithms with random projection clustering technique and also provide comparison of random number generators on Monte Carlo Simulation to solve CVRP instances with combination of random projection clustering algorithm. In the first part, we show that the random projection clustering approach improves the cost compared to the core heuristic solution. In the second part, we study the choice of the random number generators on simulation based techniques on CVRP. A Monte Carlo simulation based Clarke and Wright’s Savings (CWS) algorithm implemented and experiments conducted with five different random number generators. Results have shown the choice of random number generators affects the performance of the simulation.


       Speaker: Mervan Aksu

Advisor: Prof. Dr. Ali Devin Sezer

Zoom Link:

Meeting ID: 

Password: 

Date/Time: 01.09.2021 Perşembe saat 11:00

Place: Student Study Room, IAM (No:213) , IAM

Abstract: The classical optimal trading problem is the closure of an initial position $q_0$ in a financial asset over a time interval $[0 T]$; the trader tries to maximize an expected utility under the constraint $q_T = 0$, which is the liquidation constraint. Given that the trading takes place in a stochastic environment, the constraint $q_T=0$ may be too restrictive; the trader may want to relax this constraint or slow down/stop trading depending on price behavior. The goal of this thesis is the formulation and a study of these types of modified liquidation orders. We introduce two new parameters to the stochastic optimal control formulation of this problem: a process $I$ taking values in $\{0,1\}$ and a measurable set ${\bm S}.$ The set ${\bm S}$ prescribes when full liquidation is required and $I$ prescribes when trading takes place. We give four examples for ${\bm S}$ and $I$ which are all based on a lower bound specified for the price process. We show that the minimal supersolution of a related backward stochastic differential equation (BSDE) with a singular terminal value and with a convex driver term gives both the value function and the optimal control of the modified stochastic optimal control problem. The novelties of the BSDE arising from the modified control problem are as follows: the relaxation of the constraint $q_T = 0$ implies that the terminal value of the BSDE can take negative values; this and the convexity of the driver imply that the driver is no longer monotone and results from the currently available literature giving the minimal supersolution of this type of BSDE are not directly applicable. The same aspects of the problem imply that the BSDE can explode to $-\infty$ backward in time. To tackle these, we introduce an assumption that balances the market volume process and the permanent price impact in the model over the trading horizon. The BSDEs reduce to PDE for Markovian price processes; we also present an analysis of these PDE for a Markovian price process involving stochastic volatility.

We quantify the financial performance of our models by the percentage difference between the initial stock price and the average price at which the position is (partially) closed in the time interval $[0,T].$ We note that this difference can be divided into three pieces: one corresponding to permanent price impact ($A_1$), one corresponding to transaction/bid-ask spread costs ($A_2$) and one corresponding to random fluctuations in the price ($A_3$). $A_1$ turns out to be a linear function of $1-q_T/q_0$, the portion of the portfolio that is closed; therefore, its distribution is fully determined by that of $q_T/q_0$. We provide a numerical study of the distribution of $q_T/q_0$ and the conditional distributions of $A_2$ and $A_3$ given $q_T/q_0$ under the assumption that the price process is Brownian for a range of choices of $I$ and ${\bm S}.$


       Speaker: Kaan Çelik

       Affiliation: Cryptography

Advisor:  Assoc. Prof. Dr. Oğuz Yayla

Zoom Link:https://zoom.us/j/3122100970

Meeting ID: 3122100970

Password: 

Date/Time: 01/09/2022 10:30

Place: Student Study Room, IAM (No:213) , IAM

Abstract: With the development of technology, great developments have occurred in the healthcare area, as in every field. Over time, many solutions have been proposed for the processing of electronic health data. As a fact, there are critical factors that should be considered under these developments. This information in the field of electronic health is demanded both by some harmful organizations and people. In addition, there is an extensive market for this information. Therefore, in electronic health data systems, the privacy of the patient, the executability of the system, and its protection against attacks are milestone features. One of the methods offered to provide these security measures is the blockchain. Many theoretical and practical blockchain-based studies have been achieved for preserving electronic health data. In this thesis, we propose a blockchain based medical survey system to achieve the data integrity. The system is implemented on the Algorand blockchain and its steps are given. We also do the benchmark of the proposed system in terms of time and memory usage.


       Speaker: Güneş Batmaz Keskin

       Affiliation: 

Advisor:  Prof. Dr. Ferruh ÖZBUDAK

Co-Advisor

Zoom Link:

Meeting ID:

Password: 

Date/Time: 01.09.2022 / 10.00 

Place: Hayri Körezlioğlu Seminar Room, IAM

Abstract: This thesis presents some parts of our continuing work on Almost Perfect Nonlinear(APN) Functions. Through this thesis, first of all the significance of APN functions and couple of generation methods for APN Functions are presented. Some methods are considered in a deeper context. Finally various algorithms to obtain functions related APN functions using Python programming language have implemented and some results are analyzed.


       Speaker: İsmail Daş

       Affiliation: MSc in Financial Mathematics

Advisor:  Prof. Dr. A. Sevtap Kestel

Co-Advisor: Dr. Bilgi YILMAZ

Zoom Link:

Meeting ID:

Password: 

Date/Time: Thursday, September 1, 2022; 09:30  

Place: 207 Meeting Room, IAM

Abstract: In 2008, the whitepaper titled "Bitcoin: A Peer-to-Peer Electronic Cash System” introduced an original design of a decentralized digital currency. The innovations, described by pseuydonm Satoshi Nakamoto, have led the evolution of cryptocurrencies and have attracted significant attentions from the media and investors. The price of bitcoin has also captured the interest, and the dynamics of bitcoin have became a popular topic in academia. In this thesis, the dynamics of bitcoin price are examined in relation to hashrate and search engine metrics for the purpose of forecasting. Three periods are studied to capture the relationships in different characteristics. Cointegration relationships are examined, and according to the results, VAR and VEC models are estimated for granger causality. Unidirectional relations are found from bitcoin price to hashrate and search engine metrics. ARIMA model is used to forecast bitcoin price, the results are evaluated by some accuracy measures. 


       Speaker: Tuğba ARU

Advisor:  Doç.Dr. B.Burçak Başbuğ Erkan

Zoom Link: https://zoom.us/j/99183325553?pwd=WnlaWUIySURGQXJGOXBnSytsSmkydz09

Meeting ID: 991 8332 5553

Password: 955777

Date/Time: Friday, August 26, 2022, 11:00 

Place: Hayri Körezlioğlu Seminar Room, IAM

Abstract: This paper focuses on the effect of The 30th October Earthquake in Izmir on the number of COVID-19 cases. There have been many measures to implement social distancing and mask mandates in and around the city. However, earthquakes coinciding with a pandemic prevent their effective practice, thus increasing the proliferation of the virus. Earthquakes and other disasters make it difficult to deal with pandemics. Cities need to be more prepared for earthquakes to be better able to handle pandemics. This work combines qualitative and quantitative research based on a survey design and the analysis of the survey data. The survey results were analyzed on SPSS. We used Generalized Linear Model (GLM) to try to support the results and the model supports the earthquake has an aggravating effect on the Covid-19 pandemic.


       Speaker: Can ÖZNURLU

Affiliation: MSc in Scientific Computing

Advisor:  Prof. Dr. Ömür Uğur

Co-Advisor: Assoc. Prof. Dr. Tayfun Çimen

Zoom Link: https://zoom.us/j/99183325553?pwd=WnlaWUIySURGQXJGOXBnSytsSmkydz09

Meeting ID: 985 2117 9609

Password: 189944

Date/Time: Friday, August 26, 2022; 10:00

Place: 

Abstract: The focus of this thesis is to control the lateral-directional motion of the fighter aircraft by using integral action based Model Predictive Control (MPC) where the model is obtained by data-driven model discovery method. Dynamic Mode Decomposition with Control (DMDc) is used as a model discovery technique based only on measurement data with no modeling assumptions. The model created using this technique is used for MPC and tested against noisy conditions. In addition, performance comparison of MPC with Classical Controller is carried out. Finally, Speedgoat Unit Real-Time Target Machine®, which offers a real-time testing is used to verify the generated DMDc-MPC algorithm and understand the computational cost.
The results show that the DMDc model discovery method performs very well in noise-free situations and meets the evaluation criteria together with MPC. However, its performance decreases in the presence of measurement noise. Finally, real-time test results on Speedgoat® equipment have shown that the generated DMDc-MPC algorithm has low computational cost and can be used in systems with low computational power.


       Speaker: Burcu Özcan

       Affiliation: MSc in Financial Mathematics

Advisor:  Doç. Dr. Esma Gaygısız

Zoom Link: https://zoom.us/j/94972304428?pwd=Q3pXdkh5NnUrNTAzVGJrUVBGdTVYZz09

Meeting ID: 970 1919 8894

Password: 074421

Date/Time: Tuesday, August 23, 2022 15:30

Abstract: This project is based on Vayanos and Wang’s (2012) model. The unified model which is introduced by Vayanos and Wang’s (2012) aims to examine the effect of different imperfections on market patterns such as two measures of liquidity -including price impact (lambda), and the price reversal-, asset prices, and expected return. The unified model is tree-period model based on Grossman and Stiglitz’s canonical framework.
The objective of this project is to separately analyze how capital gain taxation and dividend shocks affect market liquidity and asset prices in discrete time by using the unified model of Vayanos and Wang (2012). In the case of capital taxation and lump-sum transfer payments, we assume that there is no capital gain tax and lump-sum transfer in Period 0. So, capital gain tax only concerns in Period 1 and Period 2, and tax rates are determined in the previous period. As the second assumption, agents receive lump-sum transfer payments together with endowment in Period 2. Both liquidity demanders and liquidity suppliers can receive lump-sum transfer payments. In the second case, we examine the effect of dividend volatility (shocks) on asset pricing and the market illiquidity measures. For that, we integrate the dividend definition in Vayanos (2000)’s paper is integrated into the Vayanos andWang’s (2010) model. We assume that there are two-period dividend payments.
As a result, we reach the conclusion that there are significant effects of both capital gain taxation and dividend payment shocks on both risk asset prices and market liquidity.


       Speaker: Buğrahan HANİKAZ

       Affiliation: MSc in Financial Mathematics

Advisor:  Doç. Dr. Esma Gaygısız

Zoom Link: https://zoom.us/j/94972304428?pwd=Q3pXdkh5NnUrNTAzVGJrUVBGdTVYZz09

Meeting ID: 

Password: 

Date/Time: Tuesday, August 23, 2022; 14:00

Place: Hayri Körezlioğlu Seminar Room

Abstract: It has been stated in various studies that the Central Bank's announcements do not only affect the short-term interest rates, but also the long-term interest rates through different channels. In this study the effect of the decision of the Central Bank of Turkey on financial markets were examined via VAR with functional shocks method proposed by Inoue and Rossi (2018) High frequency identification method was adopted to reduce the identification problem in the VAR model, impulse response functions were obtained via local projections. Two more similar VAR models were created to compare the model results. As a result of the study, monetary policy statements rarely have significant effects on the stock market in first few days, similarly, the effects on the exchange rate and implied volatility are generally insignificant, but significant results can be obtained regarding the effects on CDS. It has been observed that mostly unexpected interest rate hikes have a downward effect on CDS, while interest rate decreases have an upward effect. The effects on other financial variables cannot be interpreted as clearly as in CDS. With the established model, the short-term effects of the Central Bank's decisions on the financial markets in Turkey could only be examined for a few days, apart from the effects on first few days, some delayed effects were also observed.


       Speaker: Hilal DİNÇER

       Affiliation: MSc in Cryptography

Advisor:  Assoc. Prof. Dr. Ali Doğanaksoy

Co-Advisor: Dr. Pınar Gürkan Balıkçıoğlu

Zoom Link: https://zoom.us/j/94972304428?pwd=Q3pXdkh5NnUrNTAzVGJrUVBGdTVYZz09

Meeting ID: 949 7230 4428

Password: 748438

Date/Time: Monday, August 22, 2022; 16:00

Place: Hayri Körezlioğlu Seminar Room

Abstract: Instant messaging applications have replaced classical messaging in recent years. The fact that instant messaging applications transmit messages over the internet, therefore, being free and fast, played a major role in this rise. However, being internet-based has brought disadvantages as well as advantages. There are risks such as obtaining the message, changing the message, etc. by third parties. To avoid these risks, messages are encrypted, the sender is authenticated and their integrity is shown. However, with the developing quantum technology, it turned out that these algorithms will be broken in the near future. Now, studies are being made to make these algorithms resistant to post-quantum. In this thesis study, the key generation, key exchange, and encryption mechanisms used by the Signal Protocol in one-to-one communications, which is one of the most secure systems, are explained in detail. It is explained how open source Linphone, Xabber, Wire, and Element applications developed on the basis of Signal Protocol use Signal Protocol. In addition, in this thesis, the parameters used by Signal and Wire applications, but not specified in their documents, were obtained from open sources and added. Finally, the methods used to make the Signal Protocol quantum resistant are mentioned.


       Speaker: Özge TEKİN

       Affiliation: Phd in Financial Mathematics

Advisor:  Prof. Dr. Ömür Uğur

Co-Advisor: Prof. Dr. Rogemar S. Mamon

Zoom Link: https://zoom.us/j/94976050410?pwd=d1RXL0gxVEc5a3dIY0toMWsxVE5FZz09

Meeting ID: 949 7605 0410

Password: 457138

Date/Time: Friday, August 19, 2022, 10:00 (virtual on zoom)

Place: Hayri Körezlioğlu Seminar Room, IAM

Abstract: Economic and financial data display diverse behavior at different time intervals due to their dynamics and stochastic nature. To build explanatory models, different time periods with similar characteristics can be grouped together under a single regime. In this study, it is assumed that the states of the economy follow a homogeneous first-order continuous-time finite-state hidden Markov chain process.

We consider the valuation of European options in a three-state Markov switching extension of the Black-Scholes-Merton framework. In this context, the interest rate, drift, and volatility parameters of the underlying asset depend on the underlying market regime that switches among a finite number of states. Due to the additional source of randomness caused by the underlying Markov chain, the market is incomplete. The regime switching Esscher transform is applied to determine the equivalent martingale measure. Under this measure, the analytical formula for the regime-switching European options is derived. The option pricing procedure under this model has been studied in the literature for the two-state regime-switching framework. In this thesis, we utilize the joint density function of occupation times of the Markov chain proposed by Falzon to obtain the analytical solution for the three-state model. The calculations of the Greeks for the regime-switching European option by using the proposed method are presented.

Some of the exotic options can be represented in terms of the European options. We also consider this relationship for the barrier options and show how our method can be extended for both the valuation of regime-switching barrier options and their Greeks. The validity of the method is illustrated by presenting several examples and comparing them with the results existing in the literature.

Lastly, we consider the regime-switching guaranteed minimum maturity benefit valuation. Considering the long life of variable annuity contracts, insurance providers need to take into account both interest rate and mortality fluctuations in addition to stock market fluctuations.

We consider interest rate, mortality, and underlying fund dynamics to be switching among the states, and we propose the formulae for two different models by assuming independent filtration for the mortality component. The first model assumes that both financial and mortality parameters are regulated by the same underlying Markov chain. On the other hand, the second model assumes that the parameters of the mortality model are based on a separate second Markov chain. This study is complemented with some numerical examples to highlight the implication of our approach on pricing these contracts under a regime-switching framework.


       Speaker: Murat DEMİRCİOĞLU

       Affiliation: MS in Cryptography

Advisor:  Prof. Dr. Murat CENK

Co-Advisor: Assoc. Prof. Dr. Sedat AKLEYLEK

Zoom Link: https://zoom.us/j/98192032552?pwd=L2dGZW9rR1dXWUovWW9KS0RiQzJVdz09

Meeting ID: 981 9203 2552

Password: 181564

Date/Time: Thursday, August 04, 2022; 10:30 (virtual on zoom)

Place: Hayri Körezlioğlu Seminar Room, IAM

Abstract: The ring signature scheme has a wide range of usage areas in public-key cryptography. One of them is leaking information within a group without exposing the signer’s identity. The majority of the ring signature techniques in use, on the other hand, rely on classical crypto-systems such as RSA and ECDH, which are known to be vulnerable to Shor’s algorithm on a large-scale quantum computer. In this thesis, we propose efficient quantum-resistant ring signature schemes based on GeMSS and Gui signature algorithms. GeMSS is one of two multivariate-based signature algorithms along with Rainbow in Round 3 of Post-Quantum Cryptography Standardization Project initiated by NIST in 2016. When we compare our proposed scheme with a Rainbow-based ring signature scheme, the experimental results show that we achieve 300 times faster signature verification and almost 50 times faster signature generation as the number of users in the group increases to 50. Moreover, the proposed scheme provides %20 smaller signature sizes. Therefore, our scheme is verified to be more effective to be used.


       Speaker: Ayşegül KAHYA

       Affiliation: MS in Cryptography

Advisor:  Prof. Dr. Murat CENK

Zoom Link: https://zoom.us/j/91033592101?pwd=MCtuZzhUY2Zud3BZdkdnSmRjRU1kZz09

Meeting ID: 910 3359 2101

Password: 515012

Date/Time: Wednesday, August 03, 2022; 11:00 (virtual on zoom)

Place: Hayri Körezlioğlu Seminar Room, IAM

Abstract: For the machine learning algorithms, it was hard to work in encrypted data set because the learning algorithms tries to find patterns, and cryptographic algorithms tries to avoid creating patterns. However, the learning algorithms trained in the data set which is encrypted with homomorphic encryption can make correct inferences for the test data. The logistic regression is one of the most popular machine learning algorithms. While working logistic regression over the encrypted dataset it is needed to use some approximations on the algorithm. For this thesis, it is implemented logistic regression and approximated logistic regression algorithms. Then we encrypted a data set with a fully homomorphic encryption scheme CKKS by using Microsoft SEAL Homomorphic Encryption Library.  As a result, it is compared the successful prediction rate of logistic regression algorithm over the data set without any encryption, approximated logistic regression algorithm over the data sets without any encryption and encrypted with CKKS Scheme.


       Speaker: İrem KESKİNKURT PAKSOY

       Affiliation: PhD in Cryptography

Advisor:  Prof. Dr. Murat CENK

Place/Zoom Link: https://zoom.us/j/6073164783?pwd=RVBjekNxcTltOG05Ym1RWWpUZ0RKUT09

Meeting ID: 607 316 4783

Password: 846847

Date/Time: Thursday, July 28, 2022; 10:30 (virtual on zoom)

Place: Hayri Körezlioğlu Seminar Room, IAM

Abstract: One of the quantum-safe cryptography research areas is lattice-based cryptography. Most lattice-based schemes need efficient algorithms for multiplication in polynomial quotient rings. The fastest algorithm known for multiplication is the Number Theoretic Transform (NTT), which requires certain restrictions on the parameters of the ring, such as prime modulus. Direct NTT application is not an option for some schemes that do not comply with these restrictions, e.g., the two finalists of the PQC standardization competition, Saber and NTRU, which use a power-of-two modulus.Toom-Cookand Karatsuba are the most commonly used non-NTT multiplication algorithms. Even though a method that alters the parameters to enable NTT is proposed, the need for research on efficient non-NTT multiplication algorithms is still valid. In this thesis, we focused on developing Toeplitz Matrix-Vector Product (TMVP) based multiplication algorithms for PQC schemes. First, we propose new three- and four-way TMVP split formulas with five and seven multiplications. We choose Saber and NTRU schemes for our case study. We develop TMVP-based multiplication algorithms using the new four-way formula for the rings on which Saber and NTRU are defined. We also propose an improved version of the algorithm for Saber and present a padding method for NTRU to utilize TMVP split formulas. Moreover, we implement the proposed algorithms on ARM Cortex-M4, which NIST recommends as an evaluation platform for PQC candidates on microprocessors. We improve performance and stack memory consumption compared to all Toom implementations. We also observe that our TMVP-based algorithms are faster than NTT for three of the parameter sets of NTRU, and they reduce the stack usage for all. We integrate our codes into state-of-the-art implementations of Saber and NTRU in the literature to see the effect of our algorithm on the total performance of the schemes. For Saber, our algorithm achieves improvements up to 18.6\% in performance and up to 44.2\% in memory consumption compared to the Toom method. For all parameter sets of NTRU, we reduce stack usage between 5.9\%-20.9\% compared to Toom and 5.1\%-19.3\% compared to NTT. Moreover, we observe performance improvements between 4.4\%-17.5\% compared to Toom for all parameter sets. Except for one of the parameter sets of NTRU, our algorithms outperform the NTT method. Furthermore, we propose new formulas for non-square TMVP calculations and a new approach for deriving new TMVP split formulas using the non-square ones. The arithmetic complexity calculations and theoretical efficiency comparisons are also presented in this thesis.


       Speaker: Dilek AYDOĞAN KILIÇ

       Affiliation: PhD in Financial Mathematics

Advisor:  Prof. Dr. A. Sevtap KESTEL

Place/Zoom Link: https://zoom.us/j/96333373848?pwd=WnNuL2E4bWo4UmR0N2NsbmpJZ2pGQT09

Meeting ID: 963 3337 3848

Password: 878485

Date/Time: Tuesday, July 26, 2022; 10:00 (virtual on zoom)

Place: Hayri Körezlioğlu Seminar Room, IAM

Abstract: The aim of this thesis is to eliminate the possible weaknesses of HMMs, which is a successful statistical model that is frequently used in time series modeling. Depending on the selection of the initial parameters of the HMMs, ANN is used as a solution to the failure to reach the global maximum, and it is aimed to benefit from the classification power of this method. The hybrid model, which is developed with this motivation, is built in a way that is suitable for use in continuous data, contrary to the version generally used in the literature. In this thesis, the hybrid model, which is effective in the development of speech recognition in the literature, is reconstructed and applied to financial data. Additionally, a multivariate comparison is conducted in order to identify the effect of the other variables in the model. Therefore, bivariate and trivariate models are developed. Moreover, classical HMM and ANN applied and compared with the Hybrid model results. The applications use daily closing prices for the S\&P 500 and Nasdaq and daily EUR/USD exchange rates from 2000 to 2021. In comparison to the single HMM and ANN methods, the accuracy in forecasting is significantly increased.


       Speaker: İlayda KAYAPINAR

       Affiliation: MSc in Actuarial Science

Advisor:  Prof. Dr. A. Sevtap KESTEL

Co-Advisor: Dr. Bükre YILDIRIM KÜLEKCİ

Place/Zoom Link: https://zoom.us/j/96333373848?pwd=WnNuL2E4bWo4UmR0N2NsbmpJZ2pGQT09

Meeting ID: 963 3337 3848

Password: 878485

Date/Time: Tuesday, July 26, 2022; 10:00 (virtual on zoom)

Place: Hayri Körezlioğlu Seminar Room, IAM

Abstract: Climate components have a significant impact on the supply of agricultural goods in two ways. Firstly, the climate condition influences the efficiency of agriculture and the volume of the harvested product. Secondly, the farmers harvesting their products would prefer to sell their goods in the season when the supply is limited at a higher price. Therefore, we can say that the climate conditions can determine the price of agricultural products. Modeling the seasonal variables whose distributions are not the same and analyzing the dependence between agricultural products have great interest and importance in agricultural markets and risk theory. The copula is one of the most well-known methods for examining the dependence between variables. In this study, we employ time series analysis for Konya’s monthly adjusted weighted average prices of wheat transactions, whose clearing is conducted together with Istanbul Settlement and Custody Bank Inc., and climate components. Afterward, the adjusted spot prices against inflation is remodeled by using t-copula under the influence of climatic parameters to improve the p redictions. The main motivation behind this study is to forecast the spot wheat prices under the influence of the climate component. For this purpose, firstly, the best models are selected for the temperature, relative humidity, and precipitation, and the residuals derived from those models are used to determine the vine structure. Vine trees help us understand if there is a core climate component with the dependence structure with other variables. Then secondly, the adjusted spot wheat prices against inflation a re simulated with respect to the output of the vine structure. As a result, we see that the simulated adjusted wheat prices with t-copula give us a more accurate estimation than the predictions from the time-series analysis.


       Speaker: İlayda ÖZBABA

       Affiliation: MSc in Financial Mathematics

Advisor:  Assoc. Prof. Dr. Berna Burçak BAŞBUĞ ERKAN

Co-Advisor: Prof. Dr. A. Sevtap KESTEL

Place/Zoom Link: https://zoom.us/j/96333373848?pwd=WnNuL2E4bWo4UmR0N2NsbmpJZ2pGQT09

Meeting ID: 963 3337 3848

Password: 878485

Date/Time: Tuesday, July 19, 2022; 11:00 (virtual on zoom)

Place: Hayri Körezlioğlu Seminar Room, IAM

Abstract: Natural hazards always put human life in jeopardy and in order to protect ourselves from the effects of them, numerous disaster risk management techniques are used and it has been a critical concern for a long time. Many international institutions are dealing with beneficial programs and frameworks to decrease the detrimental effects of hazards. A recent terminology “resilience” plays an important role in this theme and resilience projects/programs are used in many continents nowadays. This thesis is oriented first to explain what resilience means and how it can be applied to financial disaster risk management. Then financial instruments that are used in financial hazard risk management are illustrated and resilience ideas in different continents are described. Main aim of the thesis is to investigate the resilience concept and its applicability in Turkiye with a focus on disaster risk reduction. The results of the analysis can provide useful recommendations for people who are interested in financial disaster risk management.


       Speaker: Rinad M. A. JUBEH

       Affiliation: MSc in Actuarial Sciences

Advisor:  Prof. Dr. A. Sevtap KESTEL

Co-Advisor: Assist. Prof. Dr. Oytun HAÇARIZ

Place/Zoom Link: https://zoom.us/j/96333373848?pwd=WnNuL2E4bWo4UmR0N2NsbmpJZ2pGQT09

Meeting number: 963 3337 3848

Password: 878485

Time: Hayri Körezlioğlu Seminar Room, IAM Tuesday, July 5, 2022; 9:00 (virtual on zoom)

Abstract: By the introduction of IFRS17, vital changes in measurements of insurance contracts are expected. This requires to assess the liabilities of insurance companies by two kinds: fulfilment cash flows and contractual service margin, in which policyholder cash flows are the nucleus of both. Hence, this thesis consists of two main parts. Firstly, we use panel data analysis to analyze policyholder cash flows in respect to Turkish returns, insurer’s cash outflows and changes in cash. Secondly, we consider a top-down modeling technique for the Turkish insurance sector. The latter uses machine learning to model, simulate, and forecast future policyholder cash flows. And compares the usage of IFRS17 with previous standards. We conclude that under IFRS17 insurers should expect their liabilities to be higher, which would change their capital structure; influencing their performance and position. This change in the liabilities of insurance companies will enhance the transparency, quality and trustfulness of the financial statements. Correspondingly, it will decrease the future variability and create homogeneity within insurance financial statements, which is the core aim of IASB in establishing IFRS17.


       Speaker: Gülden MÜLAYİM

       Affiliation: PhD in Scientific Computing

Advisor:  Prof. Dr. Bülent KARASÖZEN

Co-Advisor: Assoc. Prof. Dr. Murat UZUNCA

Place/Zoom Link: https://zoom.us/j/93108228612?pwd=WFdhRS9RZkZKTWQ3c3p1QzkvTHVQUT09

Meeting ID: 931 0822 8612

Password: 310477

Date/Time: Friday, July 1, 2022; 10:00 (virtual on zoom)

Place: Hayri Körezlioğlu Seminar Room, IAM

Abstract: In this thesis, intrusive and nonintrusive reduced-order models (ROMs) are developed for cross-diffusion systems. In the first part, we consider parameter-dependent systems with linear diffusion and cross-diffusion terms. The full-order models (FOMs) are constructed by discretizing them with finite differences in space. The resulting ordinary differential equations (ODEs) in matrix and tensor form are integrated in time with the implicit-explicit Euler (IMEX) method. The reduced bases are constructed nonintrusively with the two-level proper orthogonal decomposition (POD) approach and by applying higher-order singular value decomposition (HOSVD) to the space-time snapshots in tensor form. The reduced coefficients for new parameter values are computed using radial basis function (RBF) interpolation. The efficiency of the proposed method is illustrated through numerical experiments for two-dimensional Schnakenberg, three-dimensional Brusselator cross-diffusion equations, and predator-prey problems. In the second part of the thesis, we consider systems with nonlinear diffusion and cross-diffusion terms such as the Shigesada-Kawasaki-Teramoto (SKT) equation with Lotka-Volterra kinetics, and a tumor growth model. Finite-difference discretization of these systems in space leads to linear-quadratic ODEs. The FOMs are constructed by integrating in time with the linearly implicit Kahan's method ROMs are constructed intrusively applying POD with Galerkin projection. Exploiting the linear-quadratic structure of the ROMs, the computation of the reduced-order solutions is further accelerated by the use of POD in the tensorial framework so that the computations in the reduced system became independent of the full-order solutions. The long-term prediction capabilities of the ROMs is illustrated for one-and two-dimensional SKT equation, tumor growth problem, and predator-prey problems. Overall, the spatiotemporal patterns of cross-diffusion systems are accurately approximated by the ROMs with speedup factors of orders two and three over the full-order models.


       Speaker: Chenar Abdulla Hassan

       Affiliation: MSc in Cryptography

Advisor:  Assoc. Prof. Dr. Oğuz Yayla

Place/Zoom Link: https://zoom.us/j/3122100970

Meeting number: 3122100970

Password: No password. 

Time:  Monday, May 31, 2022 / 14:00

Abstract: The lattice-based cryptography is considered a strong candidate amongst many other proposed quantum-safe schemes for the currently deployed asymmetric cryptosystems that do not seem to stay secure when quantum computers come into play. Lattice-based algorithms possesses a time consuming operation of polynomial multiplication. As it is relatively the highest time consuming operation in lattice-based cryptosystems,  one can obtain fast polynomial multiplication by using number theoretic transform (NTT). In this thesis, we focus on and develop a radix-3 NTT polynomial multiplication and compute its computational complexity. In addition, utilizing the ring structure, we propose two parameter sets of CRYSTALS-KYBER, one of the four round three finalists in the NIST Post-Quantum Competition.


       Speaker: Ahmet Kürşat İrge

       Affiliation: MSc in Financial Mathematics

Advisor:  Seza Danışoğlu

Place/Zoom Link: https://zoom.us/j/94066158367?pwd=YVlpcm9CQk1MSGhmcGwwcXNidTBUQT09

Meeting number: 940 6615 8367

Password: 752576

Time:  Wednesday, May 11, 2022; 09:00

Abstract: This study examines whether the industry effect variables can allow investing in neglected high BM firms with the classic FSCORE method. The industry winners in the neglected firms cluster are called Underdogs. The industry effect variables examine the industry effects while the FSCORE method takes the internal picture of the high book-to-market firms. Thus, a comprehensive fundamental analysis process is established. The Generalized Method of Moment estimation explains the direction and strength of relations between the industry effects variables and future returns. The results show that the industry-winners and twelve-month market-adjusted returns have a positive relationship with statistically significant with an approximate 8% return increase for firms above the industry average. The industry winners method can separate future winners and losers in the neglected firms' cluster, so the Underdog firms produce an approximate 6% market-adjusted return increase in the twelve months. It is essential to highlight that this return increase comes from the neglected group. Consequently, the industry effect variables increased the number of investable firms by approximately 90%. The industry effect variables can separate future winners and losers in the high book-to-market firms. Moreover, the industry effects method also increased the scope and power of the FSCORE.


       Speaker: Selim Orhan

       Affiliation: MSc in Financial MathematicsMSc in Financial Mathematics

Advisor:  Seza Danışoğlu

Place/Zoom Link: https://zoom.us/j/94066158367?pwd=YVlpcm9CQk1MSGhmcGwwcXNidTBUQT09

Meeting number: 940 6615 8367

Password: 752576

Time:  May 10, 2022, 10:00 am

Abstract: Banks are considered as the marginal and sophisticated investors of financial markets. This is evident in the Haddad and Sraer (2020) study that examines the US government bond excess returns. This study extends the Haddad/Sraer analysis to the Turkish government bond market. According to the forecasting results, exposure ratio provides explanatory power over bond excess returns, especially for longer maturities. On the other hand, output gap and industrial growth present strong in-sample forecasting power for shorter-term maturities. The inclusion of macroeconomic variables into the regression along with exposure ratio increases the significance and explanatory power of exposure ratio for the explanation of bond excess returns. Output gap is the most contributive in-sample forecasting macro variable in terms of the explanation of bond excess returns. Together with output gap and exposure ratio, the inclusion of consumer price index (CPI), producer price index (PPI) or consumer confidence index improves the statistical and economic significance of in-sample regression results.


       Speaker: Süleyman Cengizci

       Affiliation: Ph.D. in Scientific Computing

Advisor:  Prof. Ömur Ugur and Prof. Tayfun E. Tezduyar

Place/Zoom Link: https://zoom.us/j/92983087322?pwd=VTB5RExYcm90cEpzeEFnTENMYmo1UT09

Meeting number: 929 8308 7322

Password: 379914

Time:  Feb 28, 2022, 10:00 Istanbul (virtual on zoom)

Abstract: For both military and civil aviation purposes, rockets, missiles, and spacecraft moving at hypersonic speeds are being utilized in recent years. While these vehicles move at speeds five times the speed of sound or more, they experience many extreme physical and chemical conditions during their flight. Because of molecular friction, such high velocities cause very high temperatures, and these high temperatures result in the excitation of the components of the gas mixture in which the vehicle moves. This situation causes various thermochemical interactions in the flow field and affects the dynamics of the flight. These interactions need to be examined accurately, for both the flight safety and the vehicle reaching the right target at the right time.

Wind tunnel experiments are both costly and insufficient to regenerate the high temperatures and shock interactions of hypersonic flights. These wind tunnel setups can also take a long time to design, test, and finally obtain the experimental data with. Therefore, computational fluid dynamics (CFD) tools are essential in analyzing the flight dynamics of hypersonic vehicles and designing them for such high speeds. Classical discretization methods need to be supplemented with stabilization and shock-capturing techniques since they suffer from spurious oscillations in simulating such high-speed flows.

In this thesis, hypersonic flows in thermochemical nonequilibrium are computationally studied. To this end, hypersonic flows of a five-species (O, N, NO, O2, N2) gas mixture around a cylinder are examined with a 17-reaction chemical model. The gas particles may be in different energy modes in hypersonic regimes due to the high temperatures: translational, rotational, vibrational, and electron-electronic. Since they have the similar time scales to reach equilibrium, the translational and rotational energy modes can be represented by one temperature, and the vibrational and electron-electronic energy modes by another. Therefore, a two-temperature chemical kinetic model is adopted.

In the computations, the compressible-flow Streamline-Upwind/Petrov--Galerkin method is employed to stabilize the finite element formulation. The stabilized formulation is supplemented with the YZβ shock-capturing to obtain good solution profiles at shocks. The nonlinear system of equations resulting from the space and time discretizations is solved with the Newton--Raphson nonlinear iterative process and ILU-preconditioned generalized minimal residual (GMRES) iterative search technique. The solvers are developed in the FEniCS environment.


       Speaker: İsa Eren Yıldırım

       Affiliation: MSc in Scientific Computing

Advisor:  Prof. Dr. Ömür UĞUR

Co Advisor : Dr. Umair Bin WAHEED

Place/Zoom Link: https://zoom.us/j/99405315740?pwd=ei9wRnVIZTUzYnlmVVhZbzZzU1FuQT09

Meeting number: 994 0531 5740

Password: 992314

Time:  Friday, February 11, 2022 / 11:00 (virtual on zoom)

Abstract: In Seismic prospecting, huge amounts of data are collected and processed to infer the structural and lithological composition of the subsurface. The key step in this procedure is velocity model building (VMB). First arrival traveltime inversion is one of the VMB tools commonly used for predicting near-surface velocity structures in seismic exploration. The underlying mathematical model describing the connection between the data (traveltimes) and the model parameters (velocity) is the eikonal equation, which is a first-order non-linear partial differential equation. Conventionally, the in version is carried out using ray-based methods or gradient-based algorithms. Though the gradient-based algorithms find the gradient that is needed to update the model parameters without requiring ray tracing, it can be computationally demanding. On the other hand, despite its robustness and efficiency ray-based methods suffer from complex regions as the ray theory relies on the high-frequency approximation. Instead of using these approaches for a traveltime inversion problem, I propose a machine learning (ML) based approach, specifically harnessing the physics informed neural net works (PINNs) exploiting the mathematical model represented by the eikonal equa tion to estimate the near-surface subsurface velocities. Training neural networks with the aid of the physics defining the underlying problem overcomes some of the challenges inherent in the traditional approaches such as requiring an acceptable a priori information, and incorrect parameter updates in the optimization. Through synthetic tests and the application of a real data, I show the reliability of the PINN based travel time inversion which can be a potential alternative tool to the traditional tomography frameworks.


       Speaker: Ertuğrul Umut Yıldırım

       Affiliation: MSc in Scientific Computing

Advisor:  Assoc. Prof.  Dr. Seza DANIŞOĞLU

Co Advisor : Assist. Prof. Dr. Guenther GLATZ

Place/Zoom Link:https://zoom.us/j/99405315740?pwd=ei9wRnVIZTUzYnlmVVhZbzZzU1FuQT09

Meeting number: 994 0531 5740

Password: 992314

Time:  Friday, February 11, 2022/10:00 (virtual on zoom)

Abstract: Computed tomography has been widely used in clinical and industrial applications as a non-destructive visualization technology. Revealing the internal structure of porous materials with the help of computed tomography scanning is at the heart of digital rock physics. The quality of computed tomography scans has a strong effect on the accuracy of the estimated physical properties of the investigated sample. X-ray exposure time is a crucial factor for scan quality. Ideally, long exposure time scans, yielding large signal-to-noise ratios, are available if physical properties are to be delineated. However, especially in micro-computed tomography applications, long exposure times constitute a problem for monitoring some physical processes that are happening quickly. To alleviate this problem, this thesis proposes a convolutional neural network approach for scan quality enhancement allowing for a reduction in X-ray exposure time while improving signal-to-noise ratio of the scanned image simultaneously. Moreover, the impact of using different loss functions, namely the mean squared error and the structural similarity index measure, on the performance of the network is analyzed. Both the visual and quantitative assessments show that the trained network greatly improves the quality of low-dose scans.


       Speaker: Abdullah Efe Gül

       Affiliation: MSc in Financial Mathematics

Advisor:  Assoc. Prof.  Dr. Seza DANIŞOĞLU

Place/Zoom Link:https://zoom.us/j/94372855962?pwd=Wm5menpQWURxNjZjRXFlN01NRjJjUT09

Meeting number: 943 7285 5962

Password: thesis

Time: Friday, February 11, 2022 / 10:00 (virtual on zoom)

Abstract: This thesis proposes two new measures of investor attention: Search Traffic (ST) and Click Per Search (CPS). These two measures as well as the commonly used Google Search Volume Index (SVI) measure are constructed by optimizing the number of keywords while using a search engine optimization. ST is measured based on financial website URLs without using any search keyword and is a direct measure of investor attention. The relationships between investor attention and stock market activities consisting of return and volatility are investigated for the Dow Jones Index (DJI) and its constituent stocks. The study provides robust evidence that attention has significant and asymmetric impact on index returns as well as excess returns. It has significant and negative influence on returns under bearish conditions while significant and positive effect during bullish conditions. Attention is also a significant driver of both index and stock volatility such that volatility increases following an increase in attention. In addition, investors respond to price reversals more quickly compared to positive index returns. Observations on CPS suggest that the more investors search for a financial keyword, the less they click on financial websites per searched keyword.

Keywords: Returns, Volatility, Investor attention, Search Engine Optimization


       Speaker: Cansu Bozkurt

       Affiliation: PhD  (or MSc) in Programme

Advisor:  Assoc.Prof.Dr. Murat CENK

Co-Advisor: Dr. Cansu BETİN ONUR

Place/Zoom Link:https://zoom.us/j/92303884017?pwd=dGFPcFNVNCttLzQ5WG4yRkNtQ1lFUT09

Meeting number:923 0388 4017

Password: 216977

Time: 10.02.2022 / 16.30

Abstract: The time period after the mid-20th century was named as information age or digital age. In that age, the world is being digitalized very fastly. The amount of data trans- ferred and processed online is increasing rapidly. As a result, data protection became an essential topic for researchers. To process or make a computation on the encrypted data  deciphering  ciphertext  first  causes  a  security  flaw. Homomorphic  encryption (HE) algorithms were designed to make computations on data without deciphering it. However, HE algorithms are able to work for a limited amount of processing steps. Fully homomorphic encryption (FHE) algorithms are developed to solve this problem. It is possible to apply any efficiently computable function on encrypted data. This thesis presents definitions, properties, applications of FHE. Some constructions of FHE schemes based on the integers are also analyzed.  Furthermore, the computational complexity of two algorithms, namely the DGHV scheme and Batch DGHVscheme (a scheme that supports encrypting and homomorphically processing a vector of plaintexts as a single ciphertext ) has been computed and their efficiency are compared based on the complexities.  While the DGHV scheme encrypts the one-bit message, the batch DGHV scheme encrypts an l-bit message vector m at a time. The primary purpose is to research which option is more efficient for encrypting l-bit messages.



       Speaker: Furkan Höçük

       Affiliation: MSc in Financial Mathematics

Place/Zoom Link: https://zoom.us/j/93230944278?pwd=WEp5S3BTMjMrNkJtaHlSNUtwMVo5UT09

Meeting number:  932 3094 4278

Password: 868382

Time: Thursday, February 10, 2022 / 14:00 (virtual on zoom)

Abstract: This empirical study compares the relative performances of the Fama-French five-factor model without foreign exchange risk and the five-factor model incorporating foreign exchange risk on capturing the cross-section of stock returns in Borsa İstanbul. We follow a similar methodology to the Fama-French's in constructing intersection portfolios and factor variables based on the several balance-sheet and income statement items of firms listed in Borsa İstanbul over July 2009 – June 2020. We propose a new proxy for foreign exchange risk on the deviations of the average stock return movements. It is known that, in emerging countries like Turkey, the tendency of firms to borrow from foreign markets where they can borrow at relatively advantageous rates and the volatile foreign exchange rates has distressed the composition of many firms' assets and liabilities in foreign currency. Our intention to suggest a proxy for foreign exchange risk is the possible effect of the composition of firms' assets and liabilities in foreign currency on average stock returns. In light of the several statistical indicators to test the predicting power, we find that both versions are good at capturing deviations in expected returns. Adding the FX risk factor to Fama-French five-factor model can slightly improve the explanatory power.  We also predicted excess returns of intersection portfolios using support vector regression method. Subsequently, we combined predictions of simple linear regression and support vector regression and found out that SVR outperforms SLR for both versions of Fama-French five-factor with and without FX risk.


       Speaker: Nazlı Ceren Demir

       Affiliation: MSc in Cryptography

Advisor:  Assist. Prof. Dr. Oğuz YAYLA

Place/Zoom Link: https://zoom.us/j/3122100970

Meeting number:Without number

Password: Without password

Time: 10.02.2022 / 16.00

Abstract: 


       Speaker: Hamdi Burak Bayrak

       Affiliation: M.Sc. in Scientific Computing

Advisor:  Assoc. Prof. Şeyda Ertekin  

Co-Advisor: Assoc. Prof. Hamdullah Yücel

Place/Zoom Link: https://us04web.zoom.us/j/7619250253?pwd=NEdVOHAwTjVVbWtpQWxlb0dYMjZjdz09

Meeting number:761 925 0253

Password: 6m6cbU

Time: February 10, 2022 10:30

Abstract:  Semi-supervised learning is a powerful approach to make use of unlabeled data to improve the performance of a deep learning model. One mostly used method of this approach is pseudo-labeling. However, pseudo-labeling, especially its originally proposed form tends to remarkably suffer from noisy training when the assigned labels are false. In order to mitigate this problem, in our work, we investigate the gradient sent to the neural network and propose a heuristic method, called competing labels, where we arrange the loss function and choose the pseudo-labels in a way that the gradient the model receives contains more than one negative element. We test our method on MNIST, Fashion-MNIST, and KMNIST datasets and show that our method has a  better generalization performance compared to the originally proposed pseudo-labeling method.



       Speaker: Özgün Ada Ceylan

       Affiliation: M.Sc. in Scientific Computing

Advisor:  Prof. Dr. Ömür UĞUR  

Place/Zoom Link: https://zoom.us/j/92373330296?pwd=N3NqRCtDaTNGaENyeGlPNzVYK1hkZz09

Meeting number: 923 7333 0296

Password: 008212

Time: Wednesday, February 09, 2022 / 10:00 (virtual on zoom)

Abstract:  Advancing technologies in distributed electrical generation and increasing amount of required electricity make electrical systems more complicated. Thus, the number of buses has in[1]creased and the size of incidence matrix for electrical power flow calculation has expanded. Hence, investigation and performance analysis on different solvers are needed. In this thesis, a performance comparison between iterative solvers for electrical power flow is studied. These solvers are Newton-Raphson, Fast-Decoupled and Newton-Krylov Spaces methods. Besides, the effect of parallel programming in Newton-Krylov Spaces Method is also investigated.


       Speaker: Giray Efe

       Affiliation: MSc in Cryptography

Advisor:  Assoc. Prof. Dr. Murat Cenk 

Place/Zoom Link: https://zoom.us/j/96559971366?pwd=Q3hteks1QlRDeGNCdTNicEdmcktGUT09

Meeting number: 965 5997 1366

Password: 577709

Time: February 09, 2022 09:50

Abstract: : Polynomial multiplication on the quotient ring is one of the most fundamental, general-purpose operations frequently used in cryptographic algorithms. Therefore, a possible improvement over a multiplication algorithm directly affects the performance of algorithms used in a cryptographic application. Well-known multiplication algorithms such as Schoolbook, Karatsuba, and Toom-Cook are dominant choices against NTT in small and ordinary input sizes. On the other hand, how these approaches are implemented under the quotient ring of polynomials,  matters. Instead of applying the reduction procedure as the final stage, using Toeplitz Matrix Product (i.e., TMVP) is a clever way to realize the modular multiplication more efficiently. Furthermore, the hybrid use of these algorithms yields more efficient results than the static choice of any single algorithm. For this purpose, we derive and analyze various constructions of multiplication and share the best possible sequences under different circumstances, and show that TMVP is a decent choice instead of classical modular polynomial multiplication approaches in cryptographic applications.



       Speaker: Berkin Aksoy

       Affiliation: MSc in Cryptography

Advisor:  Assoc. Prof. Dr. Murat Cenk 

Place/Zoom Link:  https://zoom.us/j/99299800485?pwd=eCtlVzhQcUR2aG01Z0JqZXBvWFJ4UT09

Meeting number: 992 9980 0485

Password: 185163

Time: February 09, 2022 09:00

Abstract: : Since the beginning of the National Institute of Standards and Technology (NIST), The Post-Quantum Cryptography (PQC) Standardization process, efficient implementations of lattice-based algorithms have been studied extensively. Lattice-based NIST PQC finalists use polynomial or matrix-vector multiplications on the ring with type Zq[x] / f(x). For convenient ring types, Number Theoretic Transform (NTT) can be used to perform multiplications as done in Crystals-KYBER among the finalists of the NIST PQC Standardization Process. On the other hand, if the q value of the scheme is a power of 2, as in NTRU and Saber, which are among the other lattice-based finalists, NTT can not be used explicitly. Hence multiplications are performed by the combination of Toom-Cook and Karatsuba algorithms. Recently, a novel technique called lazy interpolation has been introduced to increase the performance of Toom-Cook and Karatsuba algorithms. This thesis shows that the block recombination method is equivalent to lazy interpolation and can be used efficiently on multiplication algorithms. On the practical side, we compare different hybrid multiplication algorithms, then implement the block recombination method for Saber. Performance results are given in cycle values on general-purpose Intel processors with C implementation. Our work speeds up key generation, encapsulation, and decapsulation parts of Saber than the previous C implementations in the literature with a rate of between 10%-13%.



       Speaker: Esra Yeniaras

       Affiliation: PhD in Cryptography

Advisor:  Assoc. Prof.  Dr. Murat Cenk

Place/Zoom Link:  https://zoom.us/j/96725437392?pwd=V1dvS2lHd1AvNkZIWDdtOWFTTzN1Zz09

Meeting number: 967 2543 7392

Password: 691826

Time: January 21, 2022 14:00

Abstract: : Some of the post-quantum cryptographic protocols require polynomial multiplication in characteristic three fields, thus the efficiency of such multiplication algorithms gain more importance recently. In this thesis, we propose four new polynomial multiplication algorithms in characteristic three fields and we show that they are more efficient than the current state-of-the-art methods. We first analyze the well-known algorithms such as the schoolbook method, Karatsuba 2-way and 3-way split methods, Bernstein’s three 3-way split method, Toom-Cook-like formulas, and other recent algorithms. We realize that there are not any 4-way or 5-way split multiplication algorithms in characteristic three fields unlike the binary (characteristic two) fields which have various 4, 5, or more split versions. We then propose three different 4-way split polynomial multiplication algorithms which are derived by using the interpolation technique in F9. Furthermore, we propose a new 5-way split polynomial multiplication algorithm and then compare the arithmetic complexities and the implementation results for all of the aforementioned methods. We show that the new 4-way and 5-way split algorithms provide a 48.6% reduction in the arithmetic complexity for multiplication over F9 and a 26.8% reduction for multiplication over F3 for the input size 1280. Moreover, the new 4-way and 5-way algorithms yield faster implementation results compared   to the current state-of-the-art methods. We apply the proposed methods to NTRU Prime protocol, a key encapsulation mechanism, submitted to NIST PQC Standardization Process by Bernstein et al., which executes characteristic three polynomial multiplication in its decapsulation stage. We implement the new methods in C and observe a 26.85% speedup for stnrup653 and a 35.52% speedup for sntrup761 in the characteristic three polynomial multiplication step of the NTRU Prime decapsulation.


       Speaker: Sıtkı Can Toraman

       Affiliation: MScin Scientific Computing

Advisor:  Assoc.Prof. Dr. Hamdullah Yücel

Place/Zoom Link: https://zoom.us/j/5990725041

Meeting number: 599 072 5041

Password:

Time: Jan 6, 2022 11:00

Abstract: : Many physical phenomena such as the flow of anaircraft, or heating process, or wave ropagationare modeled mathematically by differential equations, in particular partialdifferential equations (PDEs). Analytical solutions to PDEs are often unknown orvery hard to obtain. Because of that, we simulate such systems by numerical methodssuch as finite difference, finite volume, or finite element, etc. When we want tocontrol the behavior of certain system components, such as the shape of a wingof an aircraft or an applied heat distribution, it becomes equivalent tooptimizing certain parameters of the underlying PDEs. Optimization of realworld systems in this way is called PDE-constrained optimization or optimalcontrol problems. To have a more accurate mathematical model, we employuncertain coefficients in PDEs since nature has different sources of intrinsicrandomness. In this thesis, we study a numerical investigation of a stronglyconvex and smooth tracking-type functional subject to a convection-diffusionequation with random coefficients. In spatial dimension, we use the FiniteElement Method (FEM), in probability dimension, we use the Monte Carlo (MC)method, and as an optimization method, we use the stochastic gradient (SG)method, where the true gradient is replaced by a stochastic one to minimize theexpected value over a random function. To accelerate the onvergence of the stochasticapproach, momentum terms, i.e., Polyak’s and Nesterov’s momentums, are added. Afull error analysis including Monte Carlo, finite element, and stochastic momentumgradient iteration errors are done. Numerical examples are presented toillustrate the performance of the roposed stochastic approximations in thePDEconstrained optimization setting.


       Speaker: Ahmet Şimşek

       Affiliation: M.Sc. in Cryptography

Advisor:  Oğuz Yayla 

Place/Zoom Link: https://zoom.us/j/94454255374?pwd=MTA0NFFQKzZ6VC9YMXRIU3IrbndqZz09

Meeting number: 944 5425 5374

Password:123asd

Abstract: Hyperledger was set up with the aim of being an open-source platform targeted at accelerating industry-wide collaboration hosted by The Linux Foundation for developing robust and dependable blockchain and distributed ledger-based technological platform that may be applied across several industry sectors to improve the efficiency, performance, and transactions of different business operations. Various distributed ledger frameworks and libraries have been developed for this purpose. In this thesis, the Ursa cryptographic library, which is one of the libraries being developed to offer its users with dependable, secure, user friendly and pluggable cryptographic applications to its users, has been examined and the performances of both the anonymous identity creation process and the presented cryptographic algorithms are examined.


       Speaker: Sharoy Augustine Samuel

       Affiliation: Ph.D. in Financial Mathematics

Advisor:  Ali Devin Sezer 

 To join the defense meeting please fill out the form below:

https://docs.google.com/forms/d/e/1FAIpQLSdFFS2C5tCv_hRxS_Oer43fnHDF_QtGMPWGAgVihFnfZc3TAg/viewf

To ensure the security of the meeting, the meeting info will be sent to the email addresses registered at the form above.

Date: July 2, 2021 10:30 Ankara

The  Abstract:  We study a class of nonlinear BSDEs with a superlinear driver process $f$ adapted to a filtration${\mathbb F}$ and over a random time interval $[0,S]$ where $S$ is a stopping time of ${\mathbb F}$. The filtration is assumed to support at least a $d$-dimensional Brownian motion as well as a Poisson random measure. The terminal condition $\xi$ is allowed to take the value $+\infty$, i.e., singular. Our goal is to show existence of solutions to the BSDE in this setting. We will do so by proving that the minimal supersolution to the BSDE is a solution, i.e., attains the terminal values with probability $1$. We focus on non-Markovian terminal conditions of the following form:1) $\xi = \infty \cdot {\bm 1}_{\{\tau \le S\}}$ and 2) $\xi_2 = \infty \cdot {\bm 1}_{\{ \tau >S \}}$ where $\tau$ is another stopping time.

We call a stopping time $S$ solvable with respect to a given BSDE and filtration if the BSDE has a minimal supersolution with terminal value $\infty$ at terminal time $S$. The concept of solvability plays a key role in many of the arguments. We also use the solvability concept to relax integribility conditions assumed in previous works for continuity results for BSDE with singular terminal conditions for terminal values of the form $\infty \cdot {\bm 1}_{\{\tau \le T \}}$ where $T$ is deterministic.We provide numerical examples in cases where the solution is explicitly computable and a basic application in optimal liquidation.


       Speaker: Süleyman Yıldız

       Affiliation: PhD in Scientific Computing

Advisor:  Bülent Karasözen

       Place:  https://zoom.us/j/92728802662?pwd=YkRxSUlwZ0s3MVJGZXNmdkxLRFo1QT09

Meeting number: : 927 2880 2662

Password: 361150

Date: June 18, 2021 13:30 Ankara

Abstract:  The shallow water equations (SWEs) consist of a set of two-dimensional partial differential equations (PDEs) describing a thin inviscid fluid layer flowing over the topography in a frame rotating about an arbitrary axis. SWEs are widely used in modeling large-scale atmosphere/ocean dynamics and numerical weather prediction. Highresolution simulations of the SWEs requires long time horizons over global scales, which when combined with accurate resolution in time and space makes simulations very time-consuming. While high-resolution ocean-modeling simulations are still feasible on large HPC machines, performing many query applications, such as repeated evaluations of the model over a range of parameter values, at these resolutions, is not feasible. Techniques such as reduced-order modeling produces an efficient reduced model based on existing high-resolution simulation data. In this thesis, ROMs are investigated for the rotating SWE, with constant (RSWE) and non-traditional SWE with full Coriolis force (NTSWE), and for rotating ther- mal SWE (RTSWE) while preserving their non-canonical Hamiltonian-structure, the energy and Casimirs, i.e. mass, enstrophy, vorticity, and buoyancy. Two different approaches are followed for constructing ROMs; the traditional intrusive model order reduction with Galerkin projection and the data-driven, non-intrusive ROMs. The full order models (FOM) of the SWE, which needed to construct the ROMs is obtained by discretizing the SWE in space by finite differences by preserving the skew-symmetric structure of the Poisson matrix. Applying intrusive proper orthogonal decomposition (POD) with the Galerkin projection, energy preserving ROMs are constructed for the NRSWE and RTSWE in skewgradient form. Due to nonlinear terms, the dimension of the reduced-order system scales with the dimension of the FOM. The nonlinearities in the ROM are computed by applying the discrete empirical interpolation (DEIM) method to reduce the computational cost. The computation of the reduced-order solutions is accelerated further by the use of tensor techniques. For the RSWE in linear-quadratic form, the dimension of the reduced solutions is obtained using tensor algebra without necessitating hyper-reduction techniques like the DEIM. Applying POD in a tensorial framework by exploiting matricizations of tensors, the computational cost is further reduced for the rotating SWE in linear-quadratic as well in skew-gradient form. In the data-driven, nonintrusive ROMs are learnt only from the snapshots by solving an appropriate leastsquares optimization problem in a low-dimensional subspace. Data-driven ROMs are constructed for the NTSWE and RTSWE with the operator inference (OpInf) using, (non-Markovian) and with re-projection (Markovian) dynamics, respectively. Computational challenges are discussed that arise from the optimization problem being ill-conditioned. Moreover, the non-intrusive model order reduction framework is extended to a parametric case, whereas we make use of the parameter dependency at the level of the PDE without interpolating between the reduced operators. The overall procedure of the intrusive and non-intrusive ROMs for the rotating SWEs in linear-quadratic and skew-gradient form yields a clear separation of the offline and online computational cost of the reduced solutions. The predictive capabilities of both models outside the range of the training data are shown. Both ROMs behave similarly and can accurately predict in the test and training data and capture system behavior in the prediction phase. The preservation of physical quantites in the ROMs of the SWEs such as energy (Hamiltonian), and other conserved quantities, i.e., mass, buoyancy, and total vorticity, enables that the models fit better to data and stable solutions are obtained in long-term predictions which are robust to parameter changes while exhibiting several orders of magnitude computational speedup over the FOM.


       Speaker: Umut Gölbaşı

       Affiliation: MSc in Financial Mathematics

Advisor:  A. Sevtap Kestel

       Place:  https://zoom.us/j/95369171602?pwd=ek1UUm90a3VDR1c5dWtRNGdzQkZMUT09

Meeting number: : 953 6917 1602

Password: 783645

Date: 15 March 2021, 10:00

Abstract:  Electricity generation cost and environmental effects of electricity generation continue to be among central themes in energy planning. The choice of electricity generation technology and energy source affect the environment through released greenhouse gases and other waste. United States is the world’s second-largest CO2 emitter and electricity consumer. This thesis aims to estimate the optimal capacity expansion of electric power sector in the United States for 2022-2050. We develop a fuzzy multi-objective linear program to minimize cost and environmental effects. In sensitivity analyses, we show how different policies and price evolution may alter the mix. Later on, we examine the effects of the new capacity mix and implied generation on the cost of electricity and emissions. We find that direct modeling of capacity factors give meaningful results. According to this thesis, renewable energy is expected to reach more than 1100 GW installed capacity by 2050. This reduces average cost of electricity generation by more than 70 percent and reduces CO2 emissions by more than 80 percent compared to expected end-2021 levels.


       Speaker: Burcu Aydogan

       Affiliation: PhD in Financial Mathematics

Advisor: Ömür Ugur

       Place:  https://zoom.us/j/94118246108?pwd=LzhyL255SkE1eHpQRHA2RW5vSG5aUT09

Meeting number: 941 1824 6108

Password: 316831

Date: 15 March 2021, 13:00

Abstract: In this thesis, we intend to develop optimal market making strategies in a limit order book for high-frequency trading using stochastic control approach. Firstly, we address for evolving optimal bid and ask prices where the underlying asset follows the Heston stochastic volatility model including jump components to explore the effect of the arrival of the orders. The goal of the market maker is to maximize her expected return while controlling the inventories where the remaining is charged with a liquidation cost. Two types of utility functions are considered: quadratic and exponential with a risk averse degree, respectively. Then, we take into consideration a model considering an underlying asset with jumps in stochastic volatility. We derive the optimal quotes for both models under the assumptions. For the numerical simulations, we apply finite differences and linear interpolation as well as extrapolation methods to obtain a solution of the nonlinear Hamilton-Jacobi-Bellman (HJB) equation. We discuss the influence of each parameter on the best bid and ask prices in the models and demonstrate the risk metrics including profit and loss distribution (PnL), standard deviation of PnL and Sharpe ratio which play important roles for the trader to make decisions on the strategies in high-frequency trading. Moreover, we provide the comparisons of the strategies with the existing ones. The thesis reveals that our models describe and fit the real market data better since a real data has jumps and the volatility is fluctuating in reality. As a real data application, we conduct our simulations for the developed strategies in this thesis on the high-frequency data of Borsa Istanbul (BIST). For this purpose, we first estimate the parameters of each model and then perform the numerical experiments on the optimal quotes. Our aim is to investigate the qualitative behaviour of an investor who is trading in an emerging market by our strategies in terms of the PnL, standard deviation of PnL and inventory process. Furthermore, we provide the applications on global stocks in order to see that the models are applicable, reasonable and profitable also for the developed markets. Lastly, we take account of the optimal market making models with stochastic latency in the price. The jump components are included on these models, as well. We contribute to this study by providing the numerical experiments with artificial data. Finally, the thesis ends up with a conclusion and a showcase on future research.


       Speaker: Deniz Kenan Kilic

       Affiliation: PhD in Financial Mathematics

Advisor: Ömür Ugur

       Place:  https://zoom.us/j/99252972182?pwd=Tk9Yc0d6MzlvSWhra29yYldLcGVsQT09

Meeting number: 992 5297 2182

Password: 711612

Date: 15 February 2021, 11:00

Abstract: The thesis aims to combine wavelet theory with nonlinear models, particularly neural networks, to find an appropriate time series model structure. Data like financial time series are nonstationary, noisy, and chaotic. Therefore using wavelet analysis helps for better modeling in the sense of both frequency and time. Data is divided into several components by using multiresolution analysis (MRA). Subsequently, each part is modeled by using a suitable neural network structure. In this step, the design of the model is formed according to the pattern of subseries. Then predictions of each subseries are combined. The combined prediction result is compared to the original time series’s prediction result using only a nonlinear model. Moreover, wavelets are used as an activation function for LSTM networks to form a hybrid LSTM-Wavenet model. Furthermore, the hybrid LSTM-Wavenet model is fused with MRA as a proposed method. In brief, it is studied whether using MRA and hybrid LSTM-Wavenet model decreases the loss or not for both S&P500 (∧GSPC) and NASDAQ (∧IXIC) data. Four different modeling methods are used: LSTM, LSTM+MRA, hybrid LSTM-Wavenet, hybrid LSTM-Wavenet+MRA (the proposed method). Results show that using MRA and wavelets as an activation function together decreases error values the most.


Speaker: Merve Gözde Sayın

Affiliation: MSc in Financial Mathematics

Advisor: Ceylan Yozgatligil

Place: https://zoom.us/j/91434127130?pwd=QVB1VmlRNzgrenl6b014c0ZiY3lWdz09

Meeting number: 914 3412 7130

Password: 205326

Date: 15 February 2021, 13:00

Abstract:  Estimating stock indices that reflect the market has been an essential issue for a long time. Although various models have been studied in this direction, historically, statistical methods and then various machine learning methods have to introduced artificial intelligence into our lives. Related literature shows that neural networks and tree-based models are mostly used. In this direction, in this thesis, four different models are examined. The first one is the most preferred neural network method for financial data called LSTM, and the second one is one of the most preferred tree-based models called XGBoost, and the third and the fourth models are the hybridizations of LSTM and XGBoost. Besides, these models have been applied to eight different stock market indices, and the model that gives the best results is determined according to the Mean Absolute Scaled Error (MASE) evaluation criteria.


       Speaker: Esra Günsay

       Affiliation: MSc in Cryptography

Advisor: Murat Cenk

       Place: https://zoom.us/j/91560425604?pwd=VXZqK2dOd1NYWVViVTE4UTZLR2NoUT09

       Meeting number: 915 6042 5604

       Password: 052190

       Date: 12 February 2021, 16:30

Abstract: Appropriate,  effective,  and  efficient  use  of  cryptographic protocols  contributes  tomany  novel  advances  in  real-world privacy-preserving  constructions.   One  of  the most important cryptographic protocols is zero-knowledge proofs. Zero-knowledge proofs have the utmost importance in terms of decentralized systems, especially in context of the privacy lately. In many decentralized systems, such as electronic voting,  e-cash,  e-auctions,  or anonymous credentials, zero-knowledge range proofs are used as building blocks. The main purpose of this thesis is to explain range proofs with detailed primitives and examine their applications in decentralized, so-called blockchain systems such as confidential assets, Monero, zkLedger, and Zether.In this thesis, we have examined, summarised, and compared range proofs based on zero-knowledge proofs.


       Speaker: Gizem Kara

       Affiliation: MSc in Cryptography

Advisor: Ali Doğanaksoy

Co-Advisor: Oğuz Yayla

Place: https://zoom.us/j/91560425604?pwd=VXZqK2dOd1NYWVViVTE4UTZLR2NoUT09

       Meeting number: 915 6042 5604

       Password: 052190

       Date: 12 February 2021, 17:30

Abstract: A number of arithmetization-oriented ciphers emerge for use in advanced cryptographic protocols such as secure multi-party computation (MPC), fully homomorphic encryption (FHE) and zero-knowledge proofs (ZK) in recent years. The standard block ciphers like AES and the hash functions SHA2/SHA3 are proved to be efficient in software and hardware but not optimal to use in this field for this reason, new kind of cryptographic primitives proposed. However, unlike traditional ones, there is no standard approach to design and analyze such block ciphers and the hash functions, therefore their security analysis needs to be done carefully. In 2018, StarkWare launched a public STARK-Friendly Hash (SFH) Challenge to select an efficient and secure hash function to be used within ZK-STARKs, transparent and post-quantum secure proof systems. The block cipher JARVIS is one of the first ciphers designed for STARK applications but, shortly after its publication, the cipher has been shown vulnerable to Gröbner basis attack. This master thesis aims to describe a Gröbner basis attack on new block ciphers, MiMC, GMiMCerf (SFH candidates) and the variants of JARVIS. We present the complexity of Gröbner basis attack on JARVIS-like ciphers, results from our experiments for the attack on reduced-round MiMC and a structure we found in the Gröbner basis for GMiMCerf.


       Speaker: Pınar Ongan

       Affiliation: MSc in Cryptography

Advisor: Ali Doğanaksoy   

Co-Advisor: Burcu Gülmez Temür

       Place: https://zoom.us/j/98707521401?pwd=OCtPRnV2a013QnVNUnhiTTFOQjVtUT09

       Meeting number: 987 0752 1401

       Password: 068665

       Date: 11 February 2021, 16:30

Abstract: This thesis consists of two main parts: In the first part, a study of several classes of permutation and complete permutation polynomials is given, while in the second part, a method of construction of several new classes of bent functions is described. The first part consists of the study of several classes of binomials and trinomials over finite fields. A complete list of permutation polynomials of the form f(x)=x^{(q^n-1)/(q-1)+ 1} + b x in GF(q^n)[x] is obtained for the case n=5, and a criterion on permutation polynomials of the same type is derived for the general case. Furthermore, it is shown that when q is odd, trinomials of the form f(x)= x^5 h(x^{q-1}) in GF(q^2)[x], where h(x)=x^5+x+1 never permutes GF(q^2). A method of constructing several new classes of bent functions via linear translators and permutation polynomials forms the second part of the thesis. First, a way to lift a permutation over GF(2^t) to a permutation over GF(2^m) is described, where t divides m. Then, via this method, 3-tuples of particular permutations that lead to new classes of bent functions are obtained. As a last step, the fact that none of the bent functions obtained here will be contained in Maiorana-McFarland class is proved.


Speaker: Fatih Cingöz

       Affiliation: MSc in Financial Mathematics

Advisor: Adil Oran

       Place: https://metu.webex.com/metu/j.php?MTID=m5385a4be312d961b1c561de397bb22af

Meeting number: 121 130 773

Password: Fp8vgrbq9g6

       Date: 4 February 2021, 14:00 

       Abstract:


optıon prıcıng vıa reınforcement learnıng: extensıons and

further ınsıghts on the qlbs model

Co-Advisor: Prof. Dr. Gerhard-Wilhelm Weber