Master's project where I implemented a dense net model classify sub-visual infiltration of the brainstem by analyzing only MRI images. Implemented data augmentation techniques, binary cross entropy, early stopping, and leave one out cross validation for training and evaluation.
Originally began this portion of the Prometheus project solo, and then led the subsequent members that joined. It is a neural network meant for a drone to avoid collisions with obstacles and adapt to new environments (represented in a virtual 2D grid).
This was the final project for a course where we needed to create an AI agent that would play a more complex variation of the game Pentago (one that would allow you to swap 2 quadrants of a board to add significant complexity) in a tournament against other AI players. I utilized an MCTS algorithm to create an agent that could play the game.
For this project, my task was to both work on the Neural Network half of a two-part Voting Vision system, as well as to combine both halves together (Neural Network and PCA). The voting vision system is meant to try and identify four different classes; doors, stairs, open rooms, and hallways.
Building an OS Shell in C with modular programming. Includes simulated kernel run-time environments, memory allocation, boot sequence, and a non-contiguous file system.
Coding a version of tic-tac-toe in C++ that is meant to be played across three boards at once against a computer opponent.