Capstone Final Exam Schedule

 

Final Examination Schedule

PLEASE JOIN US AS THE FOLLOWING CANDIDATES PRESENT THEIR CULMINATING WORK.

Winter 2017

 

Tuesday, February 21

Xiaolin Ma
Chair: Dr. Hazeline Asuncion
Candidate: Master of Science in Computer Science & Software Engineering
11:00 A.M.; DISC 464
Optimizing the Performance of LDA-GA Funnel Using a Parallel Computing Technique

Data provenance refer to the process of tracing and recording the origins of data and its movement between databases. LDA-GA program serves as a funnel in the data provenance reconstruction multi-level funneling model. The model has a very good precision and recall, but it is extremely time consuming. In this project, the single-threaded program is optimized with parallel computing techniques. The optimized program can work simultaneously on different machines. On each machine, the computation is carried out multi-threadedly.

By making use of multiple cores on multiple machines, the performance of LDA-GA program is improved greatly. Comparing with the single-threaded version, the multi-threaded one has a multiple times better performance.

Monday, February 27

Chris Lakin
Chair: Dr. Geethapriya Thamilarasu
Candidate: Master of Science in Cyber Security Engineering
11:30 A.M.; DISC 464
A Secure Framework for Mobile Health Applications

Recent trends indicate a steady rise in cyber attacks targeting the healthcare industry and patient data. Mobile applications in healthcare are becoming increasingly popular, and users presume these applications are inherently secure. However, a lack of security cognizance among developers and a rush to market has introduced a plethora of security vulnerabilities in mobile health applications. Our initial research showed that health-related mobile applications contain numerous vulnerabilities for attackers to potentially obtain medical data. Due to the sensitivity of the medical data, security is one of the most vital requirements for any device and application that utilizes medical data. The goal of this capstone project is to ensure secure storage and exchange of personal health information used in mobile applications while maintaining an expected level of user functionality. To this end, this project develops a security framework for mobile applications in healthcare to address common security vulnerabilities in the application development process.

Thursday, March 2

Jia Wang
Chair: Dr. William Erdly
Candidate: Master of Science in Computer Science & Software Engineering
1:30 P.M.; DISC 464
Educating Young Eyes – Shared Database and Web Portal Development

According to Prevent Blindness America, one in four school-age children has vision problems that, if left untreated, can affect learning ability, personality and adjustment in school. The child who has a near vision problem could pass the regular eye exam due to the short time of exam. In addition, the children’s eye health condition changes rapidly as they grow up. In order to find an effective way to track and detect children’s near vision problem, the Educating Young Eyes (EYE) project has been established to improve children’s eye health as well as their performance in school. The EYE project is mainly about developing interesting and kid-friendly mobile games as screening tools to help early detect children’s near vision problems and also serves as a research center to advocate the study of children’s near vision problems. This project, as part of the EYE project, proposed a shared database to support the data persistence requirements and implemented web portals for stakeholders to access and modify the data in the shared base. In particular, the shared database will store all the valuable user test data generated from mobile apps and visualize the data through web portal for parents to understand their child’s eye health, for medical practitioners to manage patient’s historical test results and prescriptions, and for researchers to further study the near vision problems and discover new solutions. Specific goals include facilitating to accommodate a variety of new assessment and vision therapy tools and games; cloud-based support for the repository; compliancy with health and student information privacy needs; and real-time as well as batch uploading of data.

Monday, March 6

Sneha Vasisht
Chair: Dr. Min Chen
Candidate: Master of Science in Computer Science & Software Engineering
10:00 A.M.; DISC 464
Video-based People Tracking

In this project, we aim to identify and track people in a given video footage. We perform face detection and recognition, and then use a block matching algorithm to continuously track each detected face. We develop a simple, intuitive, user-friendly interface that enables search by name or image that is useful and convenient for people identification and activity recognition. By plotting position data associated with each individual as a function of time, we generate a trajectory map T, which represents the movement of each person throughout the day. By overlaying the trajectories for every individual and computing the proximities of the trajectories at any given time, we build a potential interaction graph I, that identifies and records the potential interaction patterns between individuals and groups of individuals. These interaction patterns can be further analyzed to provide useful insight to improve collaboration, productivity and security.

Tuesday, March 7

Uktu Mert
Chair: Dr. Munehiro Fukuda
Candidate: Master of Science in Computer Science & Software Engineering
10:00 A.M.; DISC 464
Multi-Agent Spatial Simulation (MASS) Java Library Performance Improvement for Big Data Analysis

MASS is a parallel-computing library for multi-agent and spatial simulation over a cluster of computing nodes. The library uses two important concepts; Agent and Place. Place represents each array index of the given data set while Agent instances execute set of instructions to perform actual computation. Agents communicate with each other to exchange data and they can spawn new child Agent instances, migrate between Places, or get terminated. The library is used for conducting simulation or big data analysis. This project primarily focuses on memory efficiency of MASS Java Library for big data analysis.

Many contributors have improved this library since its initial release in 2010 and some of them made significant changes on MASS Java version during their capstone project or thesis process. Our findings show that an Agent instance uses up to 1MB of memory space and some scientific applications, such as UW Climate Analysis or Biological Network Motif, require 3 to 5 millions of Agents during their execution. This shows us that the system requires terabytes of memory which none of the computing nodes can afford. This project introduces an agent population control mechanism that restricts the number of active agents in the system and serializes new incoming agents into byte streams for later use. In addition, the library communicates with user application in each iteration cycle and transmits data. The overhead of this communication causes performance decrease in terms of both excessive memory consumption and high cpu usage. We also address this issue by implementing a practical way of doing recursive method calls along with executing spawn, kill, and migrate processes without sending data back to user application in each iteration.

Wednesday, March 8

Abirami Ramanathan
Chair: Dr. Min Chen
Candidate: Master of Science in Computer Science & Software Engineering
10:00 A.M.; DISC 464
Spatiotemporal Vehicle Tracking, Counting and Classification

Detecting and tracking vehicles from transportation surveillance videos is essential for applications ranging from traffic queue detection, volume calculation to incident and vehicle identification. However, challenges including object occlusion and shadowy condition often lead to poor detection and tracking performance. In this project, we propose a new approach to tackle these challenges. It is done in four stages: background subtraction, vehicle segmentation, shadow detection and removal, and vehicle tracking. Further, we apply the results of the above approach to perform vehicle classification and vehicle counting. Experimental results show that our method is simple, robust and effective to work on videos with occlusion issues and under various illumination/weather conditions.    

Friday, March 10

Archana Dhere
Chair: Dr. Min Chen
Candidate: Master of Science in Computer Science & Software Engineering
2:00 P.M.; DISC 464
Cloud-based Web Services for Endangered Language Analysis

Currently there are approximately 7,000 languages worldwide and one language disappears every two weeks. At this rate of extinction, only one half of these languages will continue to exist by the end of the 21st century. Language endangerment is a serious concern to which linguists have turned their attention in the last several decades. Documentation work can help to maintain, consolidate or revitalize endangered languages, and safeguard full range of uses of a given language. We aim to design, develop, and deploy web services on a cloud platform to process endangered language recordings to facilitate effective documentation and analysis. To achieve this goal, we deployed ELAN audio analysis component, time-aligned annotation, into a cloud-based web application. ELAN is an open source linguistics tool used for creation of complex annotations on audio resources. However, ELAN is a standalone application, and does not expose any libraries or services to be used in third-party applications. The deployed ELAN component can be used as a web service and integrated into any application that supports HTTP communication. We have also developed a web UI, in the Microsoft Azure cloud, for general users to access ELAN component via web browsers. With these ELAN components, users can play the audio file, view the waveform, select a specific time range of an audio file, add hierarchical tiers, and add annotations (comments/notations) based on specific requirements. In addition, all annotation related information can be saved in an EAF file on the system and viewed later for future access or sharing among users.

 

Questions? Please email Megan Jewell, CSS Graduate Advisor at mjewell@uw.edu

Masters Candidates

Abirami Ramanathan
Archana Dhere
Chris Lakin
Jia Wang
Sneha Vasisht
Uktu Mert
Xiaolin Ma


Final Examination Archive

Autumn 2016
Spring 2016
Winter 2016