Othello AI

Uses Alpha-Beta pruning and Monte Carlo Tree Search for othello AI

Othello is a classic game, involving two players who fight over control of a board. Much like a game of chess, each turn elicits more and more options for each following move, creating a decision tree that grows exponentially. In order to create a feasible and explorable decision tree, this project takes advantage of game theory algorithms that enable the "AI" to prune decision that are mathematically least optimal. It's impossible to describe these algorithms in a short summary, but I will try to avoid technical explainations and simply just link thorough explainations.

The most basic of turn-based games is mini-max where weights are determined that offer the min or max value options for each move branch. Alpha-Beta pruning builds off this by eliminating branches that are compartatively less optimal. Lastly, Monte Carlo Tree Search integrates probability, essentially finding the branch with the highest possibility of the higher moves. This offer several advantages. While alpha-beta essentially guarentees the best move to a specific depth, the depth is limited, even with pruning. Meanwhile, Monte Carlo simply chooses to follow the path with the best probability, exploring a greater depth. However, it can suffer from uncertainty and missed optimal moves. In the end, implementation is essentialy. However, I found that Monte Carlo was more effective around 60% of the time.

See the repository here.

Git Lapse

A optimization for git targeted visualization

Created for the Web and Mobile Research Lab, Git Lapse is a web app that enables visualization of Git commits, taking advantage of Myer's Diff Algorithm, which is optimized between texts that are similar. Myer's algorithm is ideal for capsulizing the difference in Git commits because commits rarely feature drastic changes, usually opting for incremental changes. Thus, combining Myer's Diff Algorithm with a simple text comparator enables quick generation of visualization.

Learn more here.

Canny Edge Detection

Can, Can, Can You Detect My Edges?

I am not proud of how long this took me, but is that not a cool picture or what? Canny Edge detection is a spin of Sobel edge detection. However, it is much better at finding definitive edges instead of simply providing context in lines. First, Canny smooths the entire image, typically done through a Gaussian filter which reduces noise and stray pixels in an image. Next, Canny finds the gradient for each pixel, done through the Sobel operator. Finally, the most important step is to trace the edges, which is done by finding all the pixels that are above a gradient threshold, determining what pixels are connected to a set, and removing all the pixels below the threshhold that do not display enough change in color.

You can learn more here and see my repository.

Nexus-Portal-Doors

A Semantic Web Solution built in MEAN Stack

The NEXUS-PORTAl-DOORS System (NPDS) is a meta-anaylsis system that uses a database to register and connect primary research papers based on content, topic, and references. By taking advantage of the semantic guidelines developed by the WWWC, NPDS is capable of creating a web of related research, linked through references and DOI links.

Learn more here

Practice Algorithms

A Fun Collection

A collection of small projects that help me explore the world of CS. When I find time and am curious about an idea that doesn't require a full repository, it falls into this collection. I do this mainly because I have a lot of fun making these.

See the repository here