Fun Class Projects

During my PhD coursework, I have done the following class projects as part of my learning experience.

Question Answering System

In our Fall 2010 Natural Language Processing class (CS6340), I worked in a class project to develop a question answering system to automatically find answers to short reading comprehension questions. In our team, we employed different strategies to identify answer snippets for different types of Why, Who, What, How, When, Where questions. Our rule based question answering system was tested on two test sets each containing 39 reading comprehension stories, and our team was ranked 1st and 3rd in the respective tests among 18 class teams that participated.

Pacman

In our Spring 2011 Artificial Intelligence class (CS6300), we implemented different AI search algorithms and learning techniques using the pacman game environment. We implemented depth-first search (DFS),  breadth-first search (BFS), Uniform-cost search, A* etc to find different locations in mazes for pacman. We also implemted adversarial search algorithms such as minimax, expectimax etc. to calculate better moves than ghost agent moves. We employed several learning techniques for pacman using value iterations, Q-learning etc. and used bayes net inference and particle filtering to determine ghost postiions. It was quite a fun learning experience.