03. Bachelor's Thesis
Permanent URI for this collection
Browse
Recent Submissions
Item Open Access ANALYSIS OF MAXIMUM MEAN DISCREPANCY GENERATIVE ADVERSARIAL NETWORKS (MMD GAN)(Nazarbayev University School of Sciences and Humanities, 2024-04-28) Kaiyrbekov, TemirlanDeep neural networks can be used to generate new data by sampling from the data distribution without explicitly defining the distribution. These nets heavy rely on optimization for efficient learning, and hence, they need mathematical guarantees for feasibility of learning. Generative Adversarial Networks (GAN) were proposed to generate images by the use of a mini-max objective function that is ”played” among two agents - a generator and a discriminator network. Later, Generative Moment Matching Networks (GMMN) were proposed to use a two-sample test instead of a discriminator network. GMMN uses Maximum Mean Discrepancy metric for distinguishing between real and generated images, but it only trains the generator network, and was implemented inefficiently. Lastly, Maximum Mean Discrepancy Generative Adversarial Networks (MMD GAN) were introduced that use adversarial kernel learning that has a mini-max objective function, efficient learning and mathematical guarantees that justify its improved performance. In this work, the mathematical reasoning behind the idea of MMD GAN was analyzed and experiments were made to tweak the parameters of the network. The loss function of MMD GAN is said to enjoy a weak topology - that MMD should tend to zero as two probability distributions converge to each other - and it will be shown empirically. Also, since the network has a loss function that is locally Lipschitz and continuous everywhere, and almost everywhere differentiable, the network was able to learn efficiently. Finally, MMD GAN with changed bandwidth parameters will be introduced that showed improved convergence with less MMD loss during training, although the loss was less smooth over epochs.Item Restricted AN EMERGENT COSMOLOGICAL MODEL FROM ASYMPTOTIC SAFETY(Nazarbayev University School of Sciences and Humanities, 2024-05-03) Zholdasbek, AknurThe modern cosmological models tell us that the early universe has gone through a phase of accelerated expansion, known as inflation, which solves the horizon and flatness problems. In this thesis, I present an inflationary cosmological model based on the Asymptotically Safe behavior of the Newton constant at Planckian energies. Unlike other models, the variability of G is given by a multiplicative coupling constant χ in the matter Lagrangian with a conserved energy-momentum tensor, and the specific functional form of χ is deduced from Asymptotically Safe gravity. In the beginning, the universe undergoes a quasi-de-Sitter phase and, then transitions to the conventional cosmological evolution after the Planck Era.Item Restricted DESIGN OF A 7-STORY HOTEL WITH A BASEMENT IN OXNARD, CA, USA(Nazarbayev University School of Engineering and Digital Sciences, 2024-04-12) Bolatkhanov, Iskander; Iskakova, Anel; Ismagulov, Nurbolat; Kaiyrkan, Assem; Khaidarova, Assemay; Satov, SerikbolsynThe following Capstone project displays a design from scratch proposed for a 7-story Hotel building with a Basement in Oxnard, California, US. Hereinafter, named as ‘élite’. The design process encompasses -four major engineering modules: Architectural/Structural, Geotechnical, Environmental, and Construction Management. Detailed calculations and design warranting structure’s vigorous stability of reinforced concrete frame in the presence of intense wind and seismic loads were conducted in the Structural part comprising 31% of the project. With a 34% share, the geotechnical component is crucial to the stability of the hotel in an earthquake-prone area because it focuses on foundation design and soil studies conducted in California. The strategies of handling Municipal Solid waste during the construction process and design life along with generation rates for the corresponding Hotel industry were revised in a detailed manner in the Environmental part of the report, correspondence of which to the project is 15%. Last but not least, the 20% of the project that is dedicated to Construction Management includes comprehensive planning, from project charter to risk management, to guarantee a smooth and efficient building process. The selection of Oxnard, a city with the required facilities and services, raises the hotel's allure to travelers. The hotel's proximity to Los Angeles and the natural wonders of the area make it an attractive business opportunity. Given Oxnard's natural characteristics as an earthquake zone and high wind speeds, the design of the climate-responsive architecture was the main concern. The project emphasizes high-quality materials and building techniques to create a firm, lifelong, and ecologically sustainable structure. The design procedures complies with the California Building Code and ASCE 7-10 standards.Item Open Access ENHANCING CREDIT CARD FRAUD DETECTION: A MULTIFACETED APPROACH USING MACHINE LEARNING AND ADAPTATION TO NON-GAUSSIAN DISTRIBUTIONS(Nazarbayev University School of Sciences and Humanities, 2024-05-01) Mustafa, AblayThis paper addresses the urgent issue of credit card fraud by using existing techniques for fraud detection, including Linear Regression, Support Vector Machines (SVM), Random Forests, Statistical Analysis, and Behavioral Analytics. My goal, as an undergraduate student, is to replicate and extend the findings presented in the ”Data mining for credit card fraud: A comparative study” paper. To implement Linear Regression, I will model the relationship between variables to identify potential fraud indicators. Support Vector Ma- chines (SVM) will involve classifying transactions into normal and fraudulent categories based on distinct patterns. Random Forests will be employed to construct an ensemble of decision trees, enhancing accuracy in detecting anomalous transactions. For Statistical Analysis, I will utilize techniques like hypothesis testing and probability distributions to analyze transaction data for irregularities. Behavioral Analytics will involve studying user behavior over time, identifying deviations from typical patterns as potential fraud signals. Considering the assumption of Gaussian distribution in existing research, I aim to expand these techniques to datasets with non-Gaussian distributions. This involves adapting algorithms to handle different statistical properties, ensuring robustness across diverse datasets. By doing so, I intend to address a gap in current literature and enhance the applicability of credit card fraud detection methods in real-world scenarios.Item Open Access WHAT WAS THE ATTITUDE OF JADID INTELLECTUALS TOWARDS THE CONQUEST OF TURKESTAN BY THE RUSSIAN EMPIRE IN THE LATE OF 19TH AND EARLY 20TH CENTURIES?(Nazarbayev University School of Sciences and Humanities, 2024-04-28) Ibragimshikov, ShokhrukhItem Restricted FLEXIBLE BIFACIAL PRINTED PEROVSKITE PHOTODETECTOR(Nazarbayev University School of Sciences and Humanities, 2024-04-29) Azamat, AidanaItem Restricted OPTICAL AND PHOTOVOLTAIC PROPERTIES OF ORGANIC SOLAR CELLS VS BULK-HETEROJUNCTION MORPHOLOGY(Nazarbayev University School of Sciences and Humanities, 2024-04-30) Asanov, NurzhanThe performance of organic bulk-heterojunction solar cells is closely tied to the morphology of their active layer, a disordered mixture of organic donor and acceptor materials, where photoexcitation and charge generation occur. In this thesis work, optical mixing principles are utilized to model the common active layer structure PM6:Y6. By controllably varying the size and shape of acceptor inclusions within a donor matrix, I investigate how changes in morphology affect the optical properties of the resulting effective medium. This finding is further utilized in transfer matrix optical simulations in combination with the Hecht equation to analyze the effect of morphology on the photovoltaic properties of organic solar cells. The reported findings can guide the optimal fabrication of similar structures by providing valuable insights into which characteristics of the blend could potentially hinder or benefit the performance of photovoltaics.Item Open Access MESOPOROUS SILICA PARTICLES FOR REMOVAL OF TOXIC, HEAVY METAL CATIONS FROM THE WATER SAMPLES(Nazarbayev University School of Sciences and Humanities, 2024-04-13) Serkul, ImanzhussipPollution of terrestrial and aquatic ecosystems with toxic and heavy metals is a major environmental concern with serious public health implications. In this study, we investigated the size effects of synthesized mesoporous silica particles (~ 50 nm, ~ 200 nm, ~ 500 nm) on the removal rate of toxic and heavy metals in tap and river water. Chemical analysis before and after adsorption experiments revealed that mesoporous silica particles can effectively remove the As, Se, V, Sn, Sr, Al, and Fe elements from water. In particular, ~200 nm SiO2 particles (surface area ~ 748.46 m2/g) showed the highest As (91.4%), Al (98.5%), Fe (72.5%) and Sr (61.9%) removal rates after 30 min. Other factors, such as water pH and contact time, were investigated as well. The results showed that fabricated mesoporous silica particles can be used for the effective removal of some toxic and heavy metals from water.Item Open Access THE EFFECT OF CONVECTION ON CORE-COLLAPSE SUPERNOVAE(Nazarbayev University School of Sciences and Humanities, 2024-04) Telman, YerassylModern Astrophysics, and the study of Core-Collapse Supernovae in particular, require an intense computational power and convenient code for different machines. A portable versatile code IDEFIX is able to deal with these issues. This thesis demonstrates a simulation of Core-Collapse Supernova with stable shock using IDEFIX. The model has an outer accreting part and inner shock. The code provides an extensive possibilities to efficiently modify the model for specific needs. Throughout the work, we obtain standing accretion shock instability (SASI) with 3D simulations, as well as observe how an outer convective shell with velocity perturbations affect the shock and SASI for different perturbation profiles. Additionally, we provide a way of analyzing those accreting perturbations which hints towards the future research of this topic.Item Restricted A FILM OF ONE’S OWN: SIGNIFICANCE OF PERSONAL SPACE AND WOMEN’S AUTONOMY IN GRETA GERWIG’S LADY BIRD (2017) AND BARBIE (2023)(Nazarbayev University School of Sciences and Humanities, 2024-05) Yerdenkyzy, AruzhanVirginia Woolf once said “A woman must have money and a room of her own if she is to write fiction” in her groundbreaking essay A Room of One’s Own (Woolf 1929, p3). Contemplating on the obstacles that women in academia are forced to overcome in order to reach the similar recognition as their male counterparts, Woolf stressed the significance of women’s acquiring their private, personal space, that allows them to be free of the responsibilities that the patriarchal world imposes onto them. Considering that she wrote this essay right at the beginning of Hollywood’s Golden Age, would it be possible for us to apply Woolf’s ideas into the new medium of cinematic world? Suppose that Greta Gerwig, Hollywood’s new name of feminist film, attempted to answer that question! As Judith Mayne wrote “the novel provides a model of the connection between art and industry later expanded by the cinema” (Mayne 1981, p29), this research examines the extent to which Greta Gerwig, as a feminist script writer and director, continued Woolf’s ideas on female authorship and developed them into a new medium. While applying close reading to the films Greta Gerwig wrote and directed herself – her debut coming of age film Lady Bird (2017) and the box-office record breaker Barbie (2023), I explored the extent to which the themes of public and private spaces, and the significance of women having personal space remain prominent in both of the works. Laura Mulvey’s essay “Visual Pleasures and Narrative Cinema” (1989) was also applied in order to evaluate the portrayal of women on screen. Gerwig’s intentional allusion to Woolf, Mulvey, and other feminist film scholars can be traced from the extensive analysis of narration and her choice of cinematic techniques as well, thus giving us opportunity to evaluate the portrayal of women on screen from the perspective of a female director, who aims to the audience of female spectators.Item Embargo BOUSSINESQ EQUATION IN ELASTIC RODS.(Nazarbayev University School of Sciences and Humanities, 2024-04-19) Zhardemova, NazerkeIn this project, the goal is to consider the equation of the transverse deflections of the rod when it is firmly connected to an elastic foundation. We study the well-posedness of the corresponding Cauchy problem, that is, existence, uniqueness and persistence properties of the solutions inherit by the initial data. Moreover, we will also find the conditions under which the solutions can extend globally in time in $L^2$ and the energy space $X$. Furthermore, we establish the persistence decay properties of the solutions in $X^s$.Item Open Access SYNTHESIS OF PS/CU MOF NANOFIBERS FOR THE ADSORPTION OF LEAD (II) CATION FROM AN AQUEOUS SOLUTION(Nazarbayev University School of Sciences and Humanities, 2024) Kuben, Ayazhan; Kuben, AyazhanWater pollution, particularly with heavy metals like lead, poses a severe threat to human health and the environment. This study explores the synthesis and application of polystyrene (PS)-supported copper metal-organic frameworks (MOFs) for efficient lead (Pb²⁺) ion removal from aqueous solutions. MOFs were synthesized using organic ligands biphenyl-4,4-dicarboxylic acid (BDC) and benzene-1,4-dicarboxylic acid (BTC) and integrated into PS nanofibers through electrospinning. The resulting nanofibers were characterized by Scanning Electron Microscopy (SEM), Energy Dispersive X-ray Spectroscopy (EDS), and Fourier Transform Infrared Spectroscopy (FTIR). Adsorption experiments demonstrated that PS/Cu-BTC nanofibers exhibited superior lead removal efficiency compared to other tested configurations, with optimal adsorption observed at 0.05g for PS/Cu-BTC nanofibers. Pure MOF powders showed even higher adsorption efficiency, highlighting the potential for improved water treatment methods. The study suggests that modifying the MOF structures and exploring new materials could further enhance lead ion removal capabilities.Item Open Access MACHINE LEARNING TECHNIQUES APPLIED TO ROBUST OPTIMAL CONTROL PROBLEMS(Nazarbayev University School of Sciences and Humanities, 2024-04-19) Zhangunissov, DilzhanThis project aims to solve the discrete time stochastic optimal control problem of evaluation of Average Value-at-Risk (AVaR) function. AVaR is an important tool in market risk management used to measure the risk. In the paper it was designed as a sequential decision model and solved by formulating an optimal control problem of minimizing the value. Brute force and Approximate Dynamic Programming (ADP) techniques were used for exact and approximate solutions respectively. Golden section search was used to solve the problem completely. The numerical experiments conducted at the end showed the effectiveness of the algorithm in evaluating the AVaR.Item Embargo HYDROTHERMAL SYNTHESIS OF TITANIA NANORODS FOR EFFICIENT PHOTOELECTROCHEMICAL WATER SPLITTING(Nazarbayev University School of Sciences and Humanities, 2024-04) Zholdasbekov, AmirPhotoelectrochemical water splitting using semiconductors such as titanium dioxide has been a highly investigated topic for several decades. Many factors can affect the photoelectrochemical activity of these materials including their phase, combination with other materials, and surface area. In this work, we optimize the synthesis procedure of titanium dioxide nanorods thin films. The morphology of these materials was investigated using SEM, and based on XRD patterns it was concluded that samples contain mixed phase rutile and anatase titania. Photoelectrochemical measurements showed that hydrothermally synthesized pristine titanium dioxide thin films can show very high and stable photocurrent densities of up to 300 μA/cm2 under AM1.5 simulated solar light. This observation opens new possibilities for designing highly efficient photoelectrochemical devices by combining TiO2 in these thin films with other semiconductors and metal nanoparticles to further enhance the activity.Item Open Access THE MEANING OF RELIGIOUS VEILING IN POST-SOVIET KAZAKHSTAN.(Nazarbayev University School of Sciences and Humanities, 2024-04-20) Tynysbekova, YelnuraFollowing the collapse of the Soviet Union, citizens in former Soviet Republics embarked on a profound "quest for identity," leading to the emergence of various religious and political groups, particularly in Central Asian states. Since the 1990s, there has been a noticeable rise of Islamic groups and an overall Islamization of society in these regions, characterized by an increase in veiled women, bearded men, mosques, and madrasas. The phenomenon of the quest for identity in former Soviet Central Asian Republics to an extent followed the flow of a broader revival of religion witnessed in Muslim-majority states like Turkey, Egypt, and Iran. The focus of this capstone project is to explore and elucidate the significance of veiling among women in post-Soviet Kazakhstan, examining the reasons behind veiling, societal and state responses, and the diverging interpretations between state agendas and individual meanings attributed to veiling. This discrepancy sheds light on the complex interplay between state policies, cultural practices, and personal interpretations of veiling.Item Restricted STATISTICAL METHODS IN NATURAL LANGUAGE PROCESSING(Nazarbayev University School of Sciences and Humanities, 2024-04-28) Nurkhan, Laiyk; Nurkhan, LaiykThis capstone project explores the application of statistical method ologies to two distinct natural language processing (NLP) tasks: machine translation between Ukrainian and Russian languages and the classifica tion of comments for hate speech detection. The study shows that the strategic integration of statistical approaches can improve performance of the machine translation and text classification problems. The imple mentation of linear regression with an added orthogonal constraint on weight vectors has resulted in higher precision scores. For the classifi cation of hate speech within textual comments, logistic regression with TF-IDF features was identified as the the most effective model in terms of AUC-ROC metric.Item Restricted LEXICAL ERRORS IN L2 RUSSIAN BY KAZAKH NATIVE SPEAKERS(Nazarbayev University School of Sciences and Humanities, 2024-04-12) Asan, AyanatThis research investigates the occurrence of lexical errors in the written language of proficient Kazakh learners of Russian. Drawing upon the Russian Learner Corpus (RLC), the study aims to categorize and analyze word choice errors made by Kazakh learners while using Russian. Specifically, it focuses on four prominent lexical mistakes based on the researcher's experiences and observations. The study emphasizes the intricate connection between Kazakh and Russian, two languages that, despite sharing some structural similarities, possess distinct vocabularies and language histories. Through an extensive analysis of a specialized corpus of Kazakh students' written texts, the research seeks to uncover patterns and factors contributing to lexical interference-based errors. Furthermore, it explores the implications of lexical errors for theories of second language acquisition and language pedagogy. Employing a mixed-methods approach encompassing qualitative and quantitative analyses, the research examines error patterns, identifies specific Kazakh translations underlying lexical mistakes, and considers contextual factors such as subject matter and text genre. The findings of this study will enhance our understanding of the development of Kazakh-Russian interlanguage and the cultural dynamics influencing it. Moreover, the research holds significance for language learners, educators, and scholars, offering valuable insights into addressing lexical errors and navigating cultural identities in multilingual contexts. To summarize, this study presents fresh perspectives on lexical interference and its impact on the language production and comprehension of proficient Kazakh learners of Russian, thereby providing meaningful contributions to academic and practical domains.Item Restricted FUNCTIONAL CHARACTERIZATION OF OSMOTIC STRESS RESPONSE IN HALOARCHAEA ISOLATED FROM THE KOBEITUZ LAKE(Nazarbayev University School of Sciences and Humanities, 2024-04-11) Bissenbayev, KudaybergenHalobacterium KBTZ01, a halophile archaea species from Kobeituz Lake, Kazakhstan, thrives in hypersaline environments. Therefore, these serve as good models for studying molecular mechanisms underlying the tolerance to high salt concentrations. This study investigates the impact of osmotic stress on these cells, induced by exposure to varying NaCl concentrations (0.6/1/1.5M) in a hypotonic growth medium at 3 and 9 hours. The experimental approach involves studying stress response by incubating cells in hypotonic media, followed by protein extraction and analysis using SDS-PAGE and mass spectrometry. The morphology and viability parameters were analyzed with confocal microscopy. Low NaCl concentrations (600) are deadly for the haloarchaea, while 1M and 1.5M NaCl can be tolerated for a certain amount of time. Proteomic findings include upregulation of certain proteins related to metabolism, biogenesis and signal processing. Significant decrease in motility was observed and was validated by downregulation of motility proteins. Some proteins indicate varying abundance levels with progression of hypotonic stress. This research contributes valuable insights into the molecular mechanisms enabling halophilic archaea to thrive in challenging environments.Item Embargo FINITE-TIME OUTPUT-FEEDBACK PASSIFICATION OF UNCERTAIN FRACTIONAL-ORDER NEURAL NETWORKS HAVING TIME-VARYING DELAYS(Nazarbayev University School of Sciences and Humanities, 2024-04-15) Serikova, TomirisThe purpose of the study was to investigate the finite-time stability and output feedback finite-time passification of fractional order uncertain neural networks with time-varying delay. We derived conditions for finite-time boundedness and finite-time passivity using linear matrix inequality, Lyapunov functional, and Schur complement Lemma we reached finite-time stability of the system. Certainly derivation and integration for fractional order calculus, that we used for neural network system, differs from conventional integer order calculus, by its hereditary characteristics. The findings have substantial implications for the design and control of complex neural network systems, paving the way for improved robustness and reliability in real-world applications.Item Open Access TRANSITION PROBABILITY OF THE PUSHTASEP WITH ONE PARAMETER(Nazarbayev University School of Sciences and Humanities, 2024-04-19) Tileukabyl, NazgulWe consider a n-particle PushTASEP, a model introduced by Alimohammadi et al. In their work they obtained the exact solution for the master equation of PushTASEP with boundary conditions. In this paper we use the technique outlined by Tracy and Widom to extend their result and find a solution which satisfies the initial condition of the model.