7264 research outputs found
Sort by
SURVEILLANCE VIDEO-TO-TEXT GENERATOR
No full textVideo-to-text generation is a relatively new field which is recently gaining a popularity as other generative models. It is particularly beneficial for surveillance to retrieve textual descriptions from CCTV cameras. Thus, this thesis aims to design an efficient video-to-text generator for surveillance. The research established that encoder-decoder neural networks are the most up-to-date technique. Specifically, pretrained CNN models and the LSTM based encoder and decoder models are the most prominent neural networks. Though the research was initially intended for surveillance videos, the largest MSR-VTT and MSVD datasets with general videos were used for training due to the absence of available surveillance dataset with captions. The models were designed in four parts: feature extraction, model training, test caption generation, and model evaluation. Video features, which were extracted using pretrained VGG16 CNN, were fed into the encoder using one LSTM layer. Then, the decoder in the form of another LSTM layer was implemented for video caption generation. In general, 12 models were trained for two datasets with various number of frames per video and vocabulary size. The best performing model, which was trained on MSVD dataset with 16 frames per video and 2000 vocabulary size, scored 12.8, 32.2, 32.9, and 44.0 on METEOR, BLEU1, ROUGE, and CIDEr evaluation metrics respectively. Therefore, MSVD dataset is the most suitable for the designed architecture. Furthermore, it was found that increasing number of frames per video was not legitimate in terms of computational resources for short videos between 10 to 30 seconds. Finally, 2000 vocabulary size is an excellent size for MSVD dataset. Though the proposed model generated captions, it performed worse than the past research in terms of evaluation metrics. This might be caused by the computational limitation, inaccurate caption datasets, and improper selection of the search algorithm
BOOSTING STUDENTS’ WELL-BEING AND ENGLISH GRAMMAR COMPETENCE THROUGH POSITIVE EDUCATION: A QUASI-EXPERIMENTAL STUDY
No full textSupporting students’ well-being and equipping them with the skills necessary to succeed in life was set as one of the 21st-century educational goals (Coleman, 2011). Positive education, unlike the traditional understanding of learning, suggests teaching regular school disciplines along with resilience and social-emotional skills (Seligman, 2011). It is believed that positive education programs integrated into school curricula can increase students’ well-being, teach them how to be perseverant, optimistic about the future, and use their strengths. Bearing this idea in mind, the purpose of this study was to analyze the effects of the positive education intervention on their subjective and psychological well-being. Also, being guided by the broaden-and-build theory, which states that an increase in positive emotions can facilitate better learning (Fredrickson, 2001), it was sought to analyze how such a program would influence their academic performance. Using a quantitative quasi-experimental pre-post-test research design, the data was collected two times from experimental and control groups consisting of 12 students in each group. The intervention was run for a week, including five lessons that aimed to strengthen students’ interpersonal relationships, sense of hope, and optimism. The analysis revealed that the positive education program had no effect on the subjective well-being (i.e., happiness) of students. However, the program demonstrated small but statistically significant gains in connectedness and academic achievement in the experimental group. It is argued that a longer duration of the intervention, as well as a more robust program, could have potentially led to more significant outcomes
SOLVING LINEAR-QUADRATIC REGULATOR PROBLEM WITH AVERAGE-VALUE-AT-RISK CRITERIA USING APPROXIMATE DYNAMIC PROGRAMMING
No full textThis master’s thesis explores the intersection of optimal control theory and risk-sensitive decision-making by addressing the finite-horizon discrete-time linear quadratic regulator (LQR) problem with a focus on the average-value-at-risk (AVaR) criteria. The study aims to mathematically formalize the LQR-AVaR problem within the dynamic programming framework and develop a computational algorithm based on approximate dynamic programming techniques to solve it. The algorithm’s effectiveness is rigorously assessed through the analysis of experiment results and plot evaluations. The experiment results indicate that the approximate dynamic programming algorithm, when applied properly, performs well for the problem, with experiments suggesting high accuracy
STM32-BASED FORMULA STUDENT RACE CAR VEHICLE CONTROL UNIT, SHUTDOWN CIRCUIT, CAN BUS COMMUNICATION
No full textThe Formula Student is an international competition aimed at enhancing engineering culture among a new generation of engineers by conducting competitions in the form of formula-type race car designs. The "NU Motorsports" club of Nazarbayev University aims to develop the first electric formula-type race car in the framework of this competition. As a senior engineer of the Grounded Low Voltage department of the club, the main goal of the current project is to develop a functional control unit that abides by all regulations of the competition. Specifically, the objective is to develop a custom VCU board that will control the inverter and be capable of running necessary algorithms pertaining to the motion of the car. Moreover, as adjacent systems of the VCU, the project goes through the implementation of the Accelerator Pedal Position Sensor (APPS), Shutdown interface circuit, and the design of a communication bus between different control units. Specifically, the given work discusses the realization of the Controller Area Network (CAN) compatible with the inverter
EATING HABITS OF NAZARBAYEV UNIVERSITY STUDENTS: HOW HAS AN ATTITUDE TOWARDS COMMENSALITY CHANGED?
No full textThis capstone project studies changes in the eating habits of Nazarbayev University students and their attitudes towards commensality, namely eating together, during their academic journey from the first to the last year. This study uses a qualitative research approach based on interview results conducted among students from different backgrounds. Eating habits reflect changes in students' social identities when adjusting to university. For example, in the first year, students are interested in integrating into the university's new society through communal meals. In contrast, they tend to strengthen their social connections and group identity in their senior year.
The practice of sharing meals helps to strengthen social capital among students. Commensality is the opportunity to create new connections and establish mutually beneficial relationships, which influences the formation of social networks and support in the university environment. Social practices in the context of the university community reflect sociocultural norms and expectations, which explain how students interact and establish their social relationships through food.
The findings may help shape cafeteria policies, organize social events, and create a supportive environment for students at the university. Additionally, the study reveals the importance of communal nutrition in forming student culture and social adaptation in the university environment. The results obtained within the project are discussed and analyzed in the context of changing perceptions and attitudes towards commensality, social relationships, and the evolution of students' eating habits
BELIEF SYSTEMS, RITUAL AT LATE IRON AGE TASBAS, KAZAKHSTAN
No full textThis capstone project examines the single burial from the Late Iron Age period in East
Kazakhstan. Earlier analysed in the dominant processualism framework, which relies on
material culture, the individual agency and lived experience in shaping cultural practices
were underestimated. Therefore, this project explores belief systems and rituals of the Wusun incorporating both processual and post-processual approaches. The multi-aspect analysis was
used including osteological analysis, grave good analysis and burial features analysis. The results of the examination of the mortuary practices at Tasbas provided the exploration of multi-aspect identities of the individual within the Wusun social and cultural contexts. With the interplay between individual’s agency, social identity, and embodied experience we were able to get a deeper understanding of the complexity of Wusun society as well as the ways in which their lived experiences were influenced by both internal and external factors
TRANSFER HYDROGENATION OF NITROGEN-CONTAINING COMPOUNDS BY A NOVEL IRON (II) CATALYST
No full textTransfer hydrogenation presents a promising alternative to direct hydrogenation, driven
by the growing demand for more cost-effective, selective, and environmentally benign catalysts. In response to this demand, the catalytic activity of iron complexes in transfer hydrogenation has been investigated. While iron has been previously employed in the transfer hydrogenation of carbonyl compounds, its role in the reduction of nitrogen-containing molecules has been relatively understudied. This unexplored avenue may offer new pathways for amine synthesis, with significant applications in diverse fields such as pharmaceuticals and agriculture. A novel iron complex has been synthesized and characterized, and its catalytic potential in the transfer hydrogenation of nitriles and other nitrogen-containing substrates is currently under investigation
PARENTAL INVOLVEMENT IN PRIMARY SCHOOL STUDENTS' ENGLISH LANGUAGE EDUCATION IN KAZAKHSTAN
No full textTo improve the quality of education, it is important that parents participate in the education of children. The primary objective of this study is to examine the different forms of parental involvement in the English language learning process of children in Kazakhstan and to identify the challenges parents encounter in offering support, utilizing a qualitative research methodology. To collect relevant information, a semi-structured interview was conducted with seven parents living in Astana. Questions were asked and analyzed in relation to Epstein's Typology of Parental Involvement. According to the results, all parents noted the importance of children's learning English for their children’s future lives. In addition, the results showed that the types of parental participatory actions in the study sample were relatively largely consistent with Epstein's Typology. To be specific, Epstein identifies six types of parental involvement. Parents engaged in four types similar to those mentioned in theory, but another two types were not found among parents. Interestingly, in addition to Epstein's Typology, two more new types of parental involvement practices were observed. Based on the findings, it is suggested that the school should help parents with resources for home learning and involve them in school activities, while teachers should communicate regularly with parents and create fun English learning events. Parents should connect with community groups for extra learning opportunitie
VALUATION OF SOME NONLINEAR FINANCIAL CONTRACTS BY FINITE ELEMENT METHOD
No full textThis thesis proposes a methodology for dealing with nonlinear financial derivative models using the finite element method (FEM). Financial engineering solutions are in high demand to mimic realistic market scenarios. Significantly, the nonlinear partial differential equations (PDE) seen in security pricing theory make it almost impossible to develop explicit solutions. Therefore, one resorts to numerical approximations. The literature contains articles dealing with nonlinear contracts using the finite difference method (FDM), which practitioners frequently use. This thesis aims to provide some computational gain in time and an accurate solution to nonlinear contracts in the derivative market. The generality of the approach is extendable to other types of American and European nonlinear contracts.
For nonlinear models, conventional FEM and Isogeometric analysis (IGA) are designed to be compared with benchmark results. The second-order P2-FEM performs better convergence properties than FDM and P1-FEM for convertible bond models. Moreover, the incorporation of an adaptive grid leads to the use of a few spatial discretizations.
Usually, PDE models seen in financial engineering consist of convection-dominated or degenerate terms. The naive approach relies on stabilization techniques, as they allow for mitigating spurious oscillations. Alternatively, we use a relatively new approach, demonstrated by IGA-NURBS-based finite element technology, where the monotonic convergence is achieved with uniform and non-uniform grids without any stabilization techniques and validated within the benchmark region. Numerical experiments were conducted among well-known conventional FEM and FDM methods. The presence of the IGA framework has showcased the classical results by using fitted curve approximation. IGA demonstrates notable results based on the linear case, where the exact solution was achieved using a lesser number of grids than those by FEM and/or FDM.
The post-processing Greek values are essential, as is the price of the contracts. The literature on computing the Greek values by FDM or finite volume methods (FVM) is vast. Specific models that consider frictionless markets may encounter challenges in accurately representing real-world scenarios. To satisfy the request of the derivative market, one shall consider the nonlinear pricing models that incorporate the specific request seen in financial derivative markets. The use of standard FDM or/and FEM leads to instability in the post-processing Greeks. In principle, a possible mitigation of such oscillations could be resolved using stabilization techniques. Employing NURBS basis functions with high compact support offers smoother Greek values, which may contribute to more reliable investment and trading strategies for hedging purposes
DETECTION OF BIOMARKERS IN TEARS WITH FIBER OPTIC BIOSENSORS
No full textTears play an important role in maintaining the normal functioning of the eyes. They
protect the eye from the harmful effects of the environment and provide hydration. Despite
the fact that tears mostly consist of water, their proteomic composition is very diverse and can
vary depending on the emotional and physical condition of a person. In addition, collecting
tears is not difficult and does not require invasive intervention. This fact makes tears an
attractive material for research. Scientists are widely working on the study of the proteomic
composition of tears and the discovery of new biomarkers of various diseases in them. The
detection of biomarkers in tears has every chance of becoming a new branch of disease
diagnosis. At the moment, a large number of biomarkers have already been found in tears and
methods for their diagnosis are being actively developed.
Biomarkers of diabetic retinopathy are among the most well-studied biomarkers in
tears. LCN1 and VEGF are the most well-known representatives of biomarkers of others. In a
normal state, the LCN1 protein is responsible for neutralizing harmful lipid molecules, but in
pathological conditions, it causes prolonged inflammation. In turn, at normal concentrations,
VEGF is responsible for the moderate development of new blood vessels, but at high
concentrations, it causes pathological neovascularization. Determining changes in the
concentrations of these proteins using fiber optic biosensors can be an effective way to
diagnose diabetic retinopathy in the early stages.
Fiber optic biosensors, such as semi-distributed interferometers have simple, fast and
low-cost sensor fabrication technology, which makes them very attractive for use in the field
of diagnostics.
In the course of this study, semi-distributed interferometric (SDI) sensors for the
detection of diabetic retinopathy biomarkers LCN1 and VEGF were developed. In the process
of optimizing sensor development, it was determined that the LCN1 sensor works most
effectively when it is functionalized at a concentration of anti-LCN1 antibodies of 8 µg/ml.
The optimal concentration for sensor functionalization for VEGF was 10 µg/m