Wiki Lang Links Explore Tool

This is a tool to compare the article links between two different language versions of a Wikipedia article. You give it one article in language A, give a second language B, and then it will find which Wikipedia article links it has in common and which ones are unique to the language they appear in. I think it’s interesting to compare what articles different language communities deem relevant to certain articles. I made this because I was bored once.

The tool

GitHub repo

Building Weighted Graph from an Image of an Environment

This project investigates using an aerial image of an environment to create a bidirectional weighted graph that can be used to navigate that environment. Originally, I only started this project only because really wanted to learn about computer vision. Then, inspired by the use of navigation graphs in games and robot pathfinding, I developed an algorithm that is meant to take an aerial image of some environment and convert it into a weighted graph. Ideally, this weighted graph could be used to efficiently create paths that could be transformed into robot machine code. This project was accepted to the Computer Vision Conference 2019. The full paper is available in the proceedings.

GPGPU Programming and P-Systems

One summer I had the opportunity to do General Purpose Graphics Processing Unit (GPGPU) Programming. We worked on various problems to determine if they were “parallelizable” and compared the runtime between solutions running on both CPUs and GPUs. All GPGPU programming was done using OpenCL. Finding some work on computing P-Systems on GPUs, which is a biologically inspired computational model that is highly parallel, I attempted to program it for OpenCL but found it very difficult at the time, so I only created the computation model for CPUs. This summer taught me a lot about parallel programming.

Using a Bilingual Dictionary to Expand Topic Model Training Data

After being interested in foreign languages and computer science for some time, I wanted to try a project related to both subjects. The purpose of this project was to try and find a way to artificially increase the amount of training data for Low Resource languages by exploiting bilingual dictionaries which have multiple translations possible between different words. While the results of the experiment showed that the method I came up with is not very promising, I got great experience working with a huge amount of data with limited hardware resources. scrapers and more to complete this project. This project was accepted to and presented at NCUR 2019 at Kennesaw state university.

Big Owen’s Data Dash

In summer 2018, I did an REU at North Carolina University called Socially-Relevant Computing and Analytics. Working in the Game2Learn lab, I worked with another undergraduate student to develop Big Owen’s data dash, an educational game meant to teach Linked Lists, which is a topic many computer science students initially struggle with. The graduate students were very helpful with development and let us conduct an educational study with one of their summer camps. Unfortunately, we anonymized the data too well and were not able to remove any data points from students who did not want their data used, so we could not share any of our results without being unethical.

Link to the game

GitHub repo