澳门赌场sis/Project Final Defense Schedule
Join us as the School of STEM master’s degree candidates present their culminating thesis and project work. 澳门赌场 schedule is updated throughout the quarter, check back for new defenses.
View previous quarter schedules
Select a master’s program to navigate to candidates:
Master of Science in Computer Science & Software Engineering
SPRING 2024
Tuesday, May 14
NAIMA NOOR
Chair: Dr. Afra Mashhadi
Candidate: Master of Science in Computer Science & Software Engineering
11:00 A.M.; Discovery Hall 464
澳门赌场sis: Fairness in Continual Federated Learning
Continual Federated Learning (CFL) is a distributed machine learning technique that enables multiple clients to collaboratively train a shared model without sharing their data, while also adapting to new classes without forgetting previously learned ones. Currently, there are limited evaluation models and metrics for measuring fairness in CFL, and ensuring fairness over time can be challenging as the system evolves. To address this, our study explores temporal fairness in CFL, examining how the fairness of the model can be influenced by the selection and participation of clients over time.
We introduce novel fairness metrics—Delta Accuracy Fairness (DAF) and Delta Forgetting Fairness (DFF)—specifically designed to ensure temporal fairness in a CFL context. Additionally, we propose a set of client selection strategies that enhance the temporal fairness of the CFL model by addressing disparities in knowledge retention. Through comprehensive analysis, we demonstrate that while no single strategy guarantees perfect temporal fairness, the Low Participation and Low Average strategies consistently outperform others in terms of stability and equity. Furthermore, our findings underscore the adaptability of the Dynamic strategy, which shows significant promise in certain tasks. 澳门赌场se insights pave the way for refining client selection strategies, enhancing CFL’s fairness, and fostering more equitable learning environments.
Wednesday, May 15
SHENYAN CAO
Chair: Dr. Munehiro Fukuda
Candidate: Master of Science in Computer Science & Software Engineering
11:00 A.M.; Discovery Hall 464
Project: An Incremental Enhancement of Agent-Based Graph Database
In the domain of big data analytics, graph Database (DB) is vital for managing complex data structures. This project focuses on enhancing an existing agent-based graph DB within the MASS Java framework. Motivated by the limitations of the existing agent-based graph DB, this project aims to enrich its capability to handle data with more detailed property information, aligning with the Property Graph Model. Through a comparative analysis of popular industry graph DBs such as Neo4j, RadisGraph, JanusGraph, and ArangoDB, this project establishes design principles focusing on the adoption of the Property Graph Model, Cypher query language, in-memory distributed graph structures, and agent utilization. 澳门赌场 project provides detailed insights into the design and implementation processes, including parsing Cypher queries to Abstract Syntax Tree (AST), planning execution strategies, and comprehensive testing to ensure system functionality and reliability. Overall, the project demonstrates the successful extension of the agent-based graph DB to handle complex and interconnected data structures, accurate execution of CREATE and MATCH cypher queries, and outlined plans for future development.
VEDANTI PAWAR
Chair: Dr. Brent Lagesse
Candidate: Master of Science in Computer Science & Software Engineering
1:15 P.M.; Join Vedanti Pawar’s Online Defense
Project: Adversarial Defense: Implementing and Evaluating Multi-Layered Strategies Against Adversarial Attacks
Deep learning (DL) has become a cornerstone in image classification tasks across various industries, notably in the development of autonomous driving systems, where it significantly enhances vehicle perception and decision-making capabilities. However, reliance on single defense mechanisms often falls short in safeguarding these models against sophisticated adversarial attacks. This research investigates the potential of combining various defense strategies to enhance the robustness of DL models, focusing on the ResNet34 and ResNet50 architectures. By employing widely-used attack methods, this study simulates real-world threats to assess whether these combined defenses can improve model accuracy and security. Testing these strategies on different network architectures across various datasets, the analysis determines the impact of each defense combination along with their computational costs. 澳门赌场 findings provide valuable insights into which strategies are most effective in different settings, guiding the development of more resilient DL systems against sophisticated attacks.
Thursday, May 16
WUBE ALEMAYEHU TUFFA
Chair: Dr. William Erdly
Candidate: Master of Science in Computer Science & Software Engineering
3:30 P.M.; Join Wube Alemayehu Tuffa’s Online Defense
Project: Transfer Learning in Neural Machine Translation for Low-Resource Languages
This project paper explores the impact of transfer learning and pre-trained models on improving Neural Machine Translation (NMT) between the low-resource language, Amharic, and the resource-rich language, English. Given the unique challenges associated with NMT for low-resource languages, this study proposes to use two innovative architectures: the Concerted Training NMT (CTNMT) and a Bert-fused NMT model, aimed at improving translation quality. 澳门赌场se models are evaluated against a conventional transformer model to determine their ability to effectively leverage pre-trained knowledge for language translation tasks.
澳门赌场 experimental approach employs the fairseq and neurST toolkit to conduct controlled experiments, with translation accuracy assessed through BLEU scores. 澳门赌场 research consolidates two smaller corpora into an expanded Amharic-English dataset, ensuring robustness and integrity for model training and evaluation while safeguarding against data leakage into the test set. 澳门赌场 CTNMT architecture utilizes rate-scheduling and dynamic switch to maximize learning from BERT through sophisticated training methodologies. Meanwhile, the Bert-fused model leverages BERT’s capabilities by embedding it within a custom-build sequence-to-sequence encoder-decoder framework.
澳门赌场 results suggest that both innovative models are effective, with the Bert-fused model achieving higher BLEU scores in both Amharic-English and English-Amharic translations compared to the baseline transformer. While the CTNMT model performed well in English-Amharic translation, it was not applicable for the opposite direction. 澳门赌场se findings highlight the potential of pre-trained models to improve the quality of Neural Machine Translation (NMT), especially for languages with limited linguistic resources. Particularly the success of these models validates the hypothesis that integrating deep bidirectional language understanding can substantially enhance translation quality, presenting a notable advancement in the field of machine translation.
Monday, May 20
YUAN MA
Chair: Dr. Munehiro Fukuda
Candidate: Master of Science in Computer Science & Software Engineering
8:45 A.M.; Join Yuan Ma’s Online Defense
Project: An Implementation of Multi-User Distributed Shared Graph
Many real-world applications such as social or biological networks can be modeled as graphs. With the increasing size of graphs, graph databases also become a popular research area. Graph databases usually require maintaining the original structure of the graph over distributed disks or preferably over distributed memory to function. Compare to those popular data-streaming tools that need to disassemble the graph into texts before processing, it’s reasonable to introduce agent-based graph computing in which we deploy agents to graphs without modifying the original shape. In this research, we introduce using Multi-Agent Spatial Simulations Library (MASS) for graph computing. Currently, most agent-based modeling (ABM) libraries including MASS focus on parallelization of ABM simulation programs. However, database systems need to accept, handle, and protect many queries from different users simultaneously, while MASS hasn’t provided users with this capability. 澳门赌场refore, this project aims at implementing a high-performance multi-user distributed shared graph and trying to add this feature to the MASS library. We have conducted research on many popular data streaming tools and distributed cache. By addressing the challenges they have on programming graph applications, we proposed and implemented a high-performance distributed shared graph structure within the MASS library. Through the performance and programmability comparison between MASS and Hazelcast (which has a distributed HashMap data structure, thus enables distributed graph construction), we demonstrated that MASS GraphPlaces has better speed when processing graph queries and at the same time offers an easier way to program graph applications such as Triangle Counting.
KENNETH TRAN
Chair: Dr. Brent Lagesse
Candidate: Master of Science in Computer Science & Software Engineering
11:00 A.M.; Join Kenneth Tran’s Online Defense
澳门赌场sis: LSTAR Framework: Lightweight Framework for Standardizing Tests for Adversarial Robustness
澳门赌场 role of neural networks in various tasks has exploded in recent years, becoming prevalent in many safety-critical applications. However, improving neural network robustness has become a challenge due to the existence of adversarial examples—imperceptible perturbations to the inputs of machine learning models that mislead classifiers into producing incorrect outputs. While there have been numerous advancements in crafting adversarial attacks and defenses, research on the basis of adversarial examples has notably lagged behind, largely due to the computational difficulty of analyzing high-dimensional spaces. This inherent difficulty has led researchers to construct models for understanding adversarial examples divergent from conventional paradigms, with some relying on commonly used frameworks while others utilize their own tailored frameworks to meet their unique needs. Consequently, replicating and building upon research in this field presents a significant challenge.
In this paper, we present a modular, lightweight framework to assist researchers in addressing these challenges by providing a comprehensive approach to evaluating machine learning models through a standardized experimentation platform. We present several potential hypotheses regarding the basis of adversarial examples and utilize our framework to verify them more robustly under complex attacks and datasets through controlled experiments. Our experimental results indicate that geometric causes directly affect the robustness of machine learning models, while statistical factors amplify the effects of adversarial attacks. 澳门赌场se findings provide a baseline for further studies to better understand the phenomenon of adversarial examples, allowing researchers to design more robust machine learning models.
Tuesday, May 21
UTKARSH DARBARI
Chair: Dr. Afra Mashhadi
Candidate: Master of Science in Computer Science & Software Engineering
8:45 A.M.; Discovery Hall 464
Project: Analyzing and Optimizing Fairness in Spatial Temporal Predictive Privacy Models
Sharing spatial-temporal data, which includes individuals’ locations and movements over time, requires careful privacy considerations to prevent re-identification from unique trajectories. This study addresses this challenge by integrating fairness principles into privacy preserving models for such data. Existing models prioritize the balance between privacy and utility (predictive accuracy), but often neglect the impact on different user groups. This can lead to potential discrimination against users based on their mobility patterns.
We propose FairMoPAE, a method that incorporates fairness metrics into the Mo-PAE model, a framework known for anonymizing spatial-temporal data. FairMoPAE leverages techniques to evaluate and improve fairness during anonymization. 澳门赌场se techniques analyze the entropy difference between original and anonymized trajectories, ensuring a more balanced trade-off between privacy and utility fairness for all users. By incorporating fairness metrics and optimizing hyperparameters, FairMoPAE aims to mitigate potential biases in the Mo-PAE model and contribute to the development of more equitable and socially responsible practices for spatial-temporal data analysis.
Wednesday, May 22
ASHISH NAGAR
Chair: Dr. David Socha
Candidate: Master of Science in Computer Science & Software Engineering
8:45 A.M.; Join Ashish Nagar’s Online Defense
Project: Empowering Mobile Learning in Resource-Constrained Communities: A Technical Exploration of Luna mHealth’s Development Process and Mobile UI Solutions
澳门赌场 rapid proliferation of mobile phones in resource-constrained settings presents a unique opportunity to leverage mobile health (mHealth) applications to improve healthcare accessibility. This research addresses the critical question of how to optimize mobile learning in areas such as the Comarca Ng?be-Buglé in Panama, where barriers to health care and education are pronounced due to low literacy levels, limited economic access to services, and the scarcity of healthcare facilities. 澳门赌场 vision is to create a tailored mobile application framework that can enhance the delivery of health education in these underserved areas, supporting interactive and multimedia content even in offline environments.
澳门赌场 imperative for this study stems from the need to bridge the digital divide and extend health education to remote and marginalized communities, where traditional healthcare delivery models fail to meet local needs. In the Comarca Ng?be-Buglé, for example, the maternal mortality rate is 58 times higher than the national average, and preventable diseases remain the leading causes of death. This region’s challenges underscore the potential impact of accessible and culturally relevant health education.
My contributions focused on a variety of modern software development techniques, such as using a component-driven architecture and focusing on efficient image file handling and content module parsing. I utilized JSON Schema for structured data interchange and employed Luna mobile app UI rendering. I followed SOLID design principles to ensure robustness and scalability. Through iterative testing, we have developed a robust foundation for the Luna mHealth framework that will, when complete, empower content creators, regardless of their technical expertise, to develop interactive mobile learning modules. This approach will help to democratize the development process and ensure that the modules are adaptable to the unique cultural and learning contexts of the target user base.
PURVA AVINASH PATIL
Chair: Dr. David Socha
Candidate: Master of Science in Computer Science & Software Engineering
11:00 A.M.; Join Purva Avinash Patil’s Online Defense
Project: Exploring the Dynamics of ‘Show and Discuss’ Sessions: Visualizing Interactive Dialogues and Rapid Feedback Cycles in a Software Engineering Course
澳门赌场 badge challenges in the Software Engineering Studio undergraduate course at the University of Washington Bothell are done in a ‘Show and Discuss’ format that provides a dynamic platform for collaborative learning. 澳门赌场 interactive format in which students show their ongoing work and engage in adaptive and emergent discussions with professors allows for rapid feedback cycles that are known to be beneficial pedagogically and in other design context such as software engineering. Our research aimed to visualize and quantify the iterative and adaptive nature of those sessions to be better able to explore the cognitive work occurring in the ‘Show and Discuss’ sessions. Using Transana software, we established a categorization system of 24 types of dialogs found in these interactions between professors and students. We then used this categorization system to code four video recordings of badge challenges for each of three different badge levels, covering a total of 7 hours and 23 minutes across the twelve sessions. We then created a visualization to help uncover interaction patterns within and across the challenge sessions. In particular, the visualizations make apparent the highly interactive and probing nature of these ‘Show and Discuss’ sessions.
SHAHRUZ MANNAN
Chair: Dr. Munehiro Fukuda
Candidate: Master of Science in Computer Science & Software Engineering
1:15 P.M.; Join Shahruz Mannan’s Online Defense
Project: Analysis and Improvement of MASS-based GIS
Geographical Information Systems (GIS) are important in several sectors due to their ability to perform the basic functions needed to capture, manage, analyze, and visualize spatial data. However, the increasing complexity and volume of the geospatial data throws a challenge to traditional GIS processing techniques. 澳门赌场refore, enhanced computational strategies should be investigated to meet demanding requirements for timely and scalable analysis. This project focuses on improving the existing integration of the Multi-Agent Spatial Simulation (MASS) library with GIS, focusing particularly on computational geometry problems used within queries. 澳门赌场 work includes a comprehensive analysis of the existing MASS-GIS system identifying the inefficiencies. It proposes strategies for improvement, implementations, and benchmarks of existing and newly identified computational geometry problems including range search, convex hull, largest empty circle, and Euclidean shortest path using both Message Passing Interface (MPI) and MASS for parallel processing, and conducting performance evaluations assessing CPU scalability, spatial scalability, and execution efficiency. 澳门赌场 findings reveal that MASS implementations have enhanced the organization and execution of spatial queries. Findings reveal that while the MASS implementations enhance the organization and execution of spatial queries, they present challenges in CPU scalability compared to traditional MPI-based systems. Notably, the MASS framework demonstrated substantial improvements in managing the existing computational geometry problems with enhanced CPU and spatial scalability compared to previous implementations in the MASS-GIS system.
Thursday, May 23
THOMAS PINKAVA
Chair: Dr. Afra Mashhadi
Candidate: Master of Science in Computer Science & Software Engineering
8:45 A.M.; Discovery Hall 464
澳门赌场sis: Deep Reinforcement Learning for Data-Agnostic Post-Training Debiasing of Black-Box Machine Learning Models
As reliance on Machine Learning systems in real-world decision-making processes grows,
ensuring these systems are free of bias against sensitive demographic groups is of increasing
importance. Existing techniques for automatically debiasing ML models generally require
access to either the models’ internal architectures, the models’ training datasets, or both. In
this paper we outline the reasons why such requirements are disadvantageous, and present
an alternative novel debiasing system that is both data- and model-agnostic. We implement
this system as a Reinforcement Learning Agent and employ it to debias four target ML
model architectures over three datasets. Our results show performance comparable to data-
and/or model-gnostic state-of-the-art debiasers.
WARREN LIU
Chair: Dr. Munehiro Fukuda
Candidate: Master of Science in Computer Science & Software Engineering
11:00 A.M.; Join Warren Liu’s Online Defense’s Online Defense
Project: Programmability and Performance Enhancement of MASS CUDA
Agent-based modeling (ABM) has proven valuable across various fields for capturing the intricacies and heterogeneity of real-world systems. However, as ABM simulations become more sophisticated and larger in scale, the need for efficient parallelization arises. Graphics Processing Units (GPUs) have emerged as a compelling alternative for parallelizing ABM simulations, offering high computational power and parallelism. In this project, we aimed to enhance the MASS CUDA library, a GPU-accelerated ABM framework, by improving its programmability and performance. We implemented essential agent functions, redesigned data structures to enable coalesced memory access, and introduced a dynamic attribute setting mechanism. 澳门赌场se enhancements led to significant improvements in programmability and performance, as demonstrated through benchmarking against previous version of MASS CUDA and a competing library, FLAME GPU 2, using five diverse applications. 澳门赌场 evaluation showcased MASS CUDA’s effectiveness in terms of programmability, performance, and scalability. 澳门赌场 improved programmability and performance of MASS CUDA enable users to focus on the modeling aspects of their simulations while harnessing the computational capabilities of GPUs. By offering a scalable and accessible framework for GPU-accelerated ABM, MASS CUDA has the potential to accelerate scientific discovery and decision-making processes in numerous fields.
MICHELLE DEA
Chair: Dr. Munehiro Fukuda
Candidate: Master of Science in Computer Science & Software Engineering
1:15 P.M.; Discovery Hall 464
Project: An Agent-Based Graph Database Benchmarking Program
Graphs can be used to represent complex relationships between different entities. 澳门赌场se are stored as edges and nodes in a graph database. This type of data makes it easier and more flexible to query connected data items, and identify insights from those relationships. Two common graph database systems are Neo4j and ArangoDB. 澳门赌场se systems store graph data differently as Neo4j is a native graph database, and ArangoDB is a multi-model database. Both database system rely on storing data in the disk, which can be slow for data retrieval when the data is not in memory. A graph database using the Multi-Agent Spatial Simulation (MASS) library is being implemented to pursue CPU and spatial scalability by leveraging distributed memory to store graph data. This project aims to provide a benchmarking protocol for performance testing of MASS compared to Neo4j and ArangoDB. This will identify the current strengths and weaknesses of MASS, and provide a standard benchmarking tool for future researchers to use. 澳门赌场 work includes using data pulled from real-world applications as the foundation of a random graph generator that takes user input regarding the topology and size of the graph that will be generated, a standard set of queries both in Cypher and AQL, a manual for testing in Neo4j, and a script for testing in ArangoDB. 澳门赌场 graph sizes used in testing are 1K nodes, 10K nodes, 20K nodes, and 30K nodes for spatial scalability evaluation via graph traversal. CPU scalability of MASS is performed on a cluster of eight computing nodes using a 10K node graph.
MINGJUN MA
Chair: Dr. Yang Peng
Candidate: Master of Science in Computer Science & Software Engineering
5:45 P.M.; Join Mingjun Ma’s Online Defense
Project: Optimizing Continuity of Applications with Parallel Machine Learning Models During Edge Server Handovers
澳门赌场 development of deep learning has continually benefited various fields, such as autonomous driving and real-time video applications. As deep learning models become increasingly complex, applications composed of deep learning models often require substantial computational resources. This computing resource can be allocated within an edge computing network. However, due to the mobility of end devices, handovers can occur when an end device moves from one signal zone to another. This transition can interrupt the inference process of deep learning applications, leading to temporary service disruptions. Frequent handovers can increase the latency of services, affecting the overall user experience. This issue is critical for real-time applications, where timely and accurate data is essential for making immediate decisions. To improve the inference quality during handover, we designed a solution and a corresponding prototype system to address this challenge. Our objective is to optimize the scheduling algorithm for non-handover and handover scenarios. For non-handover scenarios, we have optimized the system by reducing inference times. During handovers, we focus on maximizing the benefits of inference. We evaluated our design against the greedy solution and found that our approach saves more inference time and yields greater benefits than the greedy solution. 澳门赌场 results demonstrate that our solution improves inference quality in handover and non-handover scenarios.
Tuesday, May 28
ABDUL-MUIZZ IMTIAZ
Chair: Dr. Afra Mashhadi
Candidate: Master of Science in Computer Science & Software Engineering
8:45 A.M.; Join Abdul-Muizz Imtiaz’s Online Defense
Project: Detecting Toxic Emotes Across Twitch Channels
Twitch is a popular live-streaming platform with a niche language that is very different from traditional English. 澳门赌场 language used in Twitch is marked with several grammatical errors, as well as by the abundant use of emotes, which are emoji-like icons that can be used to express different emotions. This means that someone unfamiliar with the language used in Twitch may not comprehend the content of chat messages.
Pioneering research in the field of natural language processing (NLP) in Twitch proposed different techniques for sentiment analysis of Twitch comments. This project extends that work by extracting toxic emotes in a Twitch channel from chat logs, and then uses an embedding space of emotes created using Word2Vec to detect toxic emotes in other popular channels.
PARKER FORD
Chair: Dr. Kelvin Sung
Candidate: Master of Science in Computer Science & Software Engineering
11:00 A.M.; Join Parker Ford’s Online Defense
Project: Real-Time Rendering of Atmospheric Clouds
Rendering realistic clouds is an important aspect of creating believable virtual worlds. 澳门赌场 detailed shapes and complex light interactions present in clouds make this a daunting task to complete in a real-time application. Our solution, based on Schneider’s volumetric rendering and noise generation framework for low-altitude cloudscapes, supports increased realism and performance in cloud rendering. For efficient approximations of radiance measurements, we adopt Hillaire’s energy-conserving integration method for light scattering. To simulate the effect of multiple light scattering, we followed Wrenninge’s approach for computing the multi-bounce diffusion of light within a volume. To capture the details of light interreflection off microscopic water droplets, the complex behavior of Mie scattering is approximated with Jenderise and d’Eon’s phase function modeling technique. To capture the details with nominal computational cost, we introduce a temporal anti-aliasing strategy that unifies pixel and volumetric sampling. 澳门赌场 pixel area sampling integrates a blue-noise distribution with an n-rooks offset, while volumetric samples follow a stratification strategy, amortizing results over n frames.
澳门赌场 resulting system is capable of rendering scenes consisting of expansive cloudscapes well within real-time requirements, achieving frame rates between 2 and 3 milliseconds on a standard laptop. Users can adjust parameters to control various types of low-altitude cloud formations and weather conditions, with presets available for easily transitioning between settings. Our unique combination of techniques adopted in the volumetric rendering process enhances both efficiency and visual fidelity where the novel approach to volumetric temporal anti-aliasing efficiently and effectively unifies the sampling of pixel areas and volumetric intervals. Looking forward, this technique could be adapted for real-time applications such as video games or flight simulations. Further improvements could refine the cloud modeling system, incorporating procedural generation for high-altitude clouds, thus broadening the range of cloudscapes that can be represented. Additionally, recent work by Schneider has shown the potential for voxel-based cloud modeling. This modeling approach could be paired with our volumetric rendering method to further improve the appearance of the clouds.
GURKIRAT SINGH GULIANI
Chair: Dr. Min Chen
Candidate: Master of Science in Computer Science & Software Engineering
11:00 A.M.; Join Gurkirat Singh Guliani’s Online Defense
Project: MeTILDA and MultiLinguify for Language Learning
澳门赌场 capstone project focuses on the development and enhancement of language learning applications including MeTILDA ( Melodic Transcription in Language Documentation and Application) and MultiLinguify.
MeTILDA is an on-going project in our research group. It’s a web-based application for endangered language application and education. My work focused on enhancing the user experience, resolving major bugs and adopting Azure DevOps for project management. In addition, significant collaborative efforts were invested in developing comprehensive project documentation that provides detailed insights into the development process, feature implementations, and project management strategies.
澳门赌场 capstone also developed MultiLinguify, a cross platform mobile application from scratch. MultiLinguify supports the learning of different languages and enables multiple learning modes such as speaking and writing. With the target users being children aged 5-11, UI has been kept simple and intuitive. To promote self-learning, learners can draw characters and practice their pronunciations, and get real-time feedback. With scalability in mind, the current design framework accommodates scope for future enhancement like gamification, notification and additional regional languages.
Wednesday, May 29
GAGNEET SACHDEVA
Chair: Professor Mark Kochanski
Candidate: Master of Science in Computer Science & Software Engineering
1:15 P.M.; Join Gagneet Sachdeva’s Online Defense
Project: Strengthening the Teaching Tools Platform through CI/CD deployment
澳门赌场 Software Engineering Studio at the University of Washington Bothell serves as an innovative platform where students engage in live software projects, fostering real-world experience in software development. 澳门赌场 “”Teaching Tools”” website, a product of this studio, is designed as a full-stack solution that significantly enhances collaboration between students and faculty. It focuses on refining grading processes and optimising the exchange of feedback, directly addressing inefficiencies within the existing Canvas Learning Management System. This capstone project aims to further improve the educational experience at the University by identifying and resolving critical gaps in Canvas, specifically targeting issues that hinder user efficiency and complicate routine tasks.
My aim for this project is to construct a robust Continuous Integration (CI) and Continuous Deployment (CD) pipeline for the Teaching Tools website. This involves automating the build and testing processes and ensuring reliable deployment of the website to Azure cloud services, thereby extending its accessibility to a broader audience. 澳门赌场 approach incorporates CI/CD best practices leveraged through the Azure DevOps platform to enhance deployment efficiency. A strategic branching model supports this framework, maintaining a stable main branch for production releases while facilitating ongoing development in feature branches. This pipeline will not only streamline updates and feature integrations but also enable quicker releases, ensuring that enhancements and bug fixes improve user experiences in a timely fashion. By automating tests and deployments, the project reduces manual errors and increases the productivity of the development team. This allows them to focus more on creating innovative features and less on the mechanics of the deployment process. Ultimately, this infrastructure supports a dynamic educational tool, adapting quickly to the evolving needs of educators and students at the University of Washington Bothell, making it an indispensable asset in the educational landscape.
Thursday, May 30
JEFFREY MCCREA
Chair: Dr. Munehiro Fukuda
Candidate: Master of Science in Computer Science & Software Engineering
11:00 A.M.; Discovery Hall 464
Project: Enhancement of Agent Performance with Q-Learning
Graphs store data from various domains, including social networks, biological networks, transportation systems, and computer networks. As these graphs grow in size and complexity, single-machine solutions become impractical due to limitations in computational resources. Distributed graph computing addresses these challenges by leveraging multiple machines to process and analyze large-scale graphs collaboratively.
This capstone project investigates the enhancement of distributed graph computing performance in the Multi-Agent Spatial Simulation (MASS) library by integrating Q-learning for computing shortest path, closeness centrality, and betweenness centrality on distributed large-scale dynamic graphs compared to traditional and agent-based graph computing algorithms. Previous approaches in the MASS framework relied on large populations of unintelligent agents to exhaustively traverse graphs to compute solutions, making them inefficient when faced with dynamic graph data. By leveraging Q-learning and MASS’s distributed agent-based graph capabilities, we aim to optimize the decision-making processes of distributed agents, thus improving computational efficiency and accuracy.?Experimental results demonstrate that the adaptive learning mechanism of Q-learning coupled with the MASS library allows agents to dynamically adjust to changing graph structures, leading to a more robust and scalable distributed graph computing solution. This research contributes to the field of distributed systems and artificial intelligence by providing an innovative approach to enhancing multi-agent intelligence for graph computing tasks.
SHUBHAM SHANTARAM PATIL
Chair: Dr. Afra Mashhadi
Candidate: Master of Science in Computer Science & Software Engineering
1:15 P.M.; Discovery Hall 464
Project: Performance Enhanced Drowsiness Detection for Drivers Using Inception V3 and Haar Cascade.
澳门赌场 work on performance increased driver drowsiness detection system is to establish a driving system for all drivers around the globe who are at risk of falling into sleepy or drowsy states that highly compromise road safety contributing to numerous road accidents in America every year. 澳门赌场 study published by the NHTSA (National Highway Traffic Safety Administration) U.S. Department of Transportation estimates that in 2017, 91,000 police-reported crashes involved drowsy drivers. 澳门赌场se crashes led to an estimated 50,000 people injured and nearly 800 deaths. For this capstone project, we propose to create such a drowsiness detection system which will help overcome challenges in the previous studies mentioned in related works. Hybrid state-of-the-art algorithms like Inception V3 Depp CNN algorithm and Haar Cascade have been employed by the implemented systems which effectively analyze face and behavioral positions as regions of interest. To train, a large dataset consisting of 100K images containing closed/open eyes with different people (human subjects), under various lighting conditions while driving was used so that this system could become more accurate at detecting such states. As a result, we saw that our system has achieved an accuracy of 92.35% on tests. 澳门赌场 model’s performance can be evaluated by examining its Receiver Operating Characteristic curve (ROC). 澳门赌场 ROC curve generated by our model has an Area Under the Curve (AUC) value of 0.70 which implies that our system performs better than random guessing with typical ranges being between 0.7-0.8 representing good performance although may have limitations under certain contexts as well.
Friday, May 31
KARAN CHOPRA
Chair: Professor Mark Kochanski
Candidate: Master of Science in Computer Science & Software Engineering
1:15 P.M.; Join Karan Chopra’s Online Defense
Project: Applying Software Engineering To Develop Features In 澳门赌场 Teaching Tool
Canvas is an all-inclusive learning management system (LMS) that runs on the web and is intended to help with digital learning by giving institutions the ability to efficiently oversee online instruction. It gives teachers the resources they need to design, present, and evaluate online courses, and it gives students the chance to engage in classes, monitor their skill growth, and get feedback on their academic achievement. Canvas is a key platform in the field of digital education, with features designed to meet the various needs of educational institutions and their learning communities. Even with its extensive capability, there is still room for improvement to raise the bar for both teacher and student productivity and learning outcomes.
Seeing this room for growth, this capstone project takes the form of an effort to create a full-stack application that runs in the browser. Through feature enhancements and a comparative study of current functionalities, this project aims to deliver new features that follow strict software engineering principles. 澳门赌场 development of multiple stand-alone features for the Teaching Tools program is the main goal of this project. 澳门赌场se features include importing and exporting quizzes, refactoring previous code to make it more comprehensible, redesigning the architecture of the application, and working on the UI development of new features. It is intended for these elements to be smoothly integrated with Canvas, enhancing the University of Washington Bothell’s virtual learning environment. Additionally, acting as a mentor to undergraduate students and fostering communication and collaboration are key components of this initiative. 澳门赌场 project intends to give all users a more dynamic and engaging learning experience by directly integrating these advancements into Canvas. In addition, this project is dedicated to applying good software engineering principles, guaranteeing an efficient system’s design and execution that places a premium on a clear and captivating user interface. With this project, the hope is to raise the bar for digital learning platforms and create a setting where education and technology meet to improve the educational experience.
Master of Science in Cybersecurity Engineering
SPRING 2024
Wednesday, May 15
TIMOTHY LUM
Chair: Dr. William Erdly
Candidate: Master of Science in Cybersecurity Engineering
3:30 P.M.; Discovery Hall 464
Project: 澳门赌场 Howdu (“How do I…?”) Project: A Knowledge Management System to Support Cybersecurity Implementation
Knowledge capture and distribution has been a perennial human endeavor since prehistory. In the modern era, the internet has facilitated an exponential growth in the volume of information available, however it presents a sub-ideal resource in providing consistent, structured instruction for how best to implement effective cyber defenses.
In this capstone, we evaluate the technical pressures and missing linkages that have prevented guidance from being presented to mitigate cyberattacks. We then build a cyber defense knowledge repository that allows users to create a cumulative snapshot of their experiences and insights.
In limited testing, Howdu facilitated a 61% speedup of an arbitrary and partially obfuscated task for those performing it (Practitioners). It further altered the subjective perception of this task from “”Confusing””, “”Frustrating””, and “”Ambiguous”” to a more positive outlook of “”Easy””, “”Comprehensible””, and “”Fun””. 澳门赌场se initial results suggest the application’s ability to improve network defense by aiding defender efficiency, decreasing stress, and reducing burnout.
Future Works include an integration of the system for grading information – called the Trust Index – and implementation of a system for translating knowledge articles across languages – called the Gnosetta. Other long term goals include containerization of the application and reductions in third-party service reliance.
Monday, May 20
KEVIN HUANG
Chair: Dr. Marc Dupuis
Candidate: Master of Science in Cybersecurity Engineering
1:15 P.M.; Join Kevin Huang’s Online Defense
Project: Targeted Cybersecurity Awareness and Training Programs for Two Asiatic Minority Groups in the U.S.
Cybersecurity education provides the knowledge and skills necessary for consumers to navigate the digital world safely. This knowledge is not equally accessible for all consumers. 澳门赌场re is currently a lack of cybersecurity education available for Asian minority groups in the U.S. Language and cultural barriers make it difficult for consumers in these groups to receive the knowledge that they need. 澳门赌场 goal of our project was to create effective cybersecurity training programs for the Chinese and Vietnamese minority groups. We accomplished this by first creating education materials through incorporating the latest research and guidelines for password security and social engineering prevention. We then found participants to attend our training sessions through the help of nonprofit organizations. After conducting the training sessions, our results confirmed that there is a demand from the Chinese and Vietnamese minority groups for cybersecurity education. Our analysis also showed that the training sessions made an impact on the participants’ intentions to improve their password hygiene and be more vigilant against social engineering attempts. 澳门赌场 insight gained from this project can be used to expand the research and development of cybersecurity education to different Asian minority groups, additional cybersecurity topics, and additional cities across the U.S.
Thursday, May 23
CHALERMWAT PUAPOLTHEP
Chair: Dr. Brent Lagesse
Candidate: Master of Science in Cybersecurity Engineering
3:30 P.M.; Join Chalermwat Puapolthep’s Online Defense
Project: Privacy-Preserving Source Code Similarity Detection
Reusing or sharing the source code could potentially violate license agreements and lead to lawsuits for an organization. 澳门赌场se infringements can be identified using source code similarity detection tools. However, existing methods lack sufficient protection for the privacy of the source code. Due to privacy concerns, the software creator might be hesitant to share their source code for investigation. 澳门赌场refore, we propose a source code similarity detection technique that preserves the privacy of the source code by implementing a normalizing process to standardize the source code before hashing it. Our comparison algorithm employs fuzzy hashing, which enables us to evaluate similarity based on the hash values. Our proof-of-concept application efficiently detects Type I and II code clones with minor time trade-offs. To verify the trade-off, we experimented to compare the time usage with and without the normalizing process. Additionally, the accuracy of Type II code clone detection is investigated. 澳门赌场 result showed an improvement in precision and recall, with an average time taken of 0.0005 seconds longer than the process without normalizing.
Friday, May 24
CHHEANG DUONG
Chair: Dr. Brent Lagesse
Candidate: Master of Science in Cybersecurity Engineering
8:45 A.M.; Join Chheang Duong’s Online Defense
Project: Generating Synthetic Data and Evaluating its Privacy
澳门赌场 field of Generative AI continues to grow in popularity in recent years from organizations seeking new ways to leverage synthetic data to optimize their processes. This technology is especially useful in areas like healthcare, where sharing data for research needs to balance between keeping information private and making sure the data is still useful. 澳门赌场 deployment of generative AI for data anonymization prior to public release is being increasingly considered by organizations. Synthetic data offers the benefit of preserving privacy while maintaining the utility of the data for research purposes. However, the assumption that synthetic data does not contain real data can create a false sense of security if the underlying machine learning models are inadequately configured. Thus, organizations must be able to assess the privacy risk of the synthetic data.
This study sought to establish the foundations for a comprehensive framework that evaluates the privacy risks associated with synthetic data. Our approach incorporates previous works on synthetic data generation and privacy assessment into a single workflow consisting of three phases. First, our Data Synthesis module generates synthetic data utilizing Conditional Tabular Generative Adversarial Networks (CTGAN) or Differentially Private CTGAN (DP-CTGAN). Next, our Privacy Attack module runs Membership Inference Attacks (MIAs) against the synthetic data to identify potential privacy leaks. As one of the three types of privacy attack strategies, MIAs aim to reverse engineer the machine learning model and deduce if a data sample is part of the original data. Finally, our Privacy Evaluation module analyzes the Data Utility and Privacy Defense of the synthetic datasets by translating the results of the Privacy Attack module into quantitative metrics.
澳门赌场 results of our testing provide valuable insights for future research in optimizing and adapting our workflow. For example, current literature on synthetic data privacy recommends leveraging Differential Privacy to improve privacy risk, but they did not validate data utility. Low data utility would essentially render the synthetic data useless. Our results found that while Differential Privacy significantly reduced the probability of a successful privacy attack, data utility also decreased considerably. This finding supports the need for further optimization of our Data Synthesis module, assisting users in generating statistically similar synthetic data. In the future, we may be able to build a robust framework that would allow organizations to confidently generate and release synthetic datasets for public consumption without compromising individual privacy.
SAJA ALSULAMI
Chair: Dr. Marc Dupuis
Candidate: Master of Science in Cybersecurity Engineering
11:00 A.M.; Join Saja Alsulami’s Online Defense
澳门赌场sis: A Study on the Effectiveness of Education and Fear Appeal to Prevent Spear Phishing of Online Users
Spear phishing attack is considered one of the most elaborate attacks in social engineering. It presupposes that an attacker designs a scam to obtain the personal information of specific users from their social media accounts. It involves a preliminary analysis of targeted users and their online behaviors needed to persuade them that a malicious link or attachment is sent by a trusted person. This attack implies that human beings are the weakest link within a security system; their vulnerabilities could be exploited. 澳门赌场 most detrimental consequences following spear phishing attacks are financial losses, network compromises, loss of login credentials, and malware installation.
This quantitative study aims to examine the impact of education and fear appeals on users’ knowledge and abilities to identify spear phishing attacks. Three interventions were implemented: an educational intervention, a fear appeal intervention, and a combined educational-fear appeal intervention. 澳门赌场 control group was used for comparison purposes. This study was conducted as an online experiment with 726 participants, and they were assigned randomly into four groups; after interventions, there was a test to evaluate their knowledge and abilities to identify spear phishing attacks. 澳门赌场 test was administered to compare the efficacy of every intervention group (educational, fear appeal, and combined educational-fear appeal) to the control group. 澳门赌场 experiment findings revealed no statistically significant differences in the mean test for these four groups. 澳门赌场 study results indicate further research is needed to develop an effective intervention program that would considerably enhance users’ knowledge of spear phishing attacks and their resilience to them.
Tuesday, May 28
ANTHONY JESUS BUSTAMANATE SUAREZ
Chair: Dr. Brent Lagesse
Candidate: Master of Science in Cybersecurity Engineering
3:30 P.M.; Discovery Hall 464
澳门赌场sis: Advancing Deep Packet Inspection in SDNs: A Comparative Analysis of P4 and OpenFlow Programmability
This thesis undertakes a critical examination of Deep Packet Inspection (DPI) capabilities within
Software-Defined Networking (SDN) frameworks, emphasizing the comparative efficacy of P4 programming language against the conventional OpenFlow protocol.
OpenFlow, while foundational in SDN’s evolution, exhibits notable constraints in DPI’s domain,
primarily due to its limited packet inspection depth, confined largely to the transport layer (Layer 4).
In contrast, this research advocates for the adoption of P4 for its unparalleled flexibility and programmability, extending DPI functionalities to the application layer (Layer 7), thereby addressing and potentially surpassing OpenFlow’s limitations.
Employing a methodical approach, this study harnesses Open vSwitch and BMv2 (Behavioral Model
version 2) switches to simulate real-world network scenarios. 澳门赌场se simulations facilitate a head-to-
head comparison of OpenFlow and P4 in executing DPI tasks, particularly focusing on HTTP and SQL protocols — common vectors for network threats. Through a comprehensive suite of protocols including OpenFlow, gRPC (Google Remote Procedure Call), and P4Runtime, the research crafts a robust DPI framework, further complemented by a custom-developed controller designed for the BMv2 and P4 ecosystem. This controller’s introduction marks a significant step in demonstrating DPI’s operational viability and efficiency within an SDN environment, leveraging application-layer traffic management. 澳门赌场 research culminates providing three different implementations to do Deep Packet Inspection within SDN benchmarking each of them to measure their advantages and disadvantages. With these implementations and benchmarking, we not only aim to validate P4’s superiority over OpenFlow in managing DPI tasks but we also seek to dynamically adapt packet-processing techniques to the ever-evolving landscape of network threats. By advancing SDN functionalities beyond traditional layer boundaries, this thesis contributes significantly to the discourse on network security, management, and optimization, paving the way for future innovations in increasingly complex network environments.
Master of Science in Electrical Engineering
SPRING 2024
Tuesday, May 14
CHERYL KUNG
Chair: Dr. Walter Charczenko
Candidate: Master of Science in Electrical Engineering Engineering
3:30 P.M.; Discovery Hall 464
澳门赌场sis: Low-Cost 3-D Printed Helical Antenna with Dielectric Support
Satellite-based internet connection requires high directivity, millimeter wave phased array antennas to be able to receive and transmit signals effectively. Phased array antennas for millimeter waves have historically been very expensive to manufacture. Exploring low-cost methods for manufacturing high directivity antennas may bring down the costs of these systems, allowing more equitable access to internet.
Helical antennas are a type of high directivity antenna that can be used for these purposes. However, helical antennas are difficult to manufacture and scale due to its three dimensional (3-D) shape of the helix conductor. New 3-D printing technology allows the creation of a dielectric support for the helical antenna element. This adds mechanical rigidity to the antenna and is feasible for high volume manufacturing at a lower cost.
This thesis explores the design of a low-cost helical antenna using a 3-D printed dielectric core for mechanical support. 澳门赌场 research in this thesis concludes that it is possible to design a helical antenna using low-cost dielectric materials with high relative permittivity at microwave frequencies. As a proof of principle, a 5 GHz helical antenna embedded in a solid dielectric was designed and modeled using electromagnetic field simulation software. At 5 GHz, the software simulations can be compared to helical antennas that are manufactured on conventional 3-D printers and commonly used resin dielectrics. 澳门赌场 conclusion and results of the computer simulations show that helical antennas with dielectric support will radiate in the axial mode with high directivity and circular polarization.