IT3105 - Artificial Intelligence Programming

This repo contains 2 of the 3 practical projects done in the course Artificial Intelligence Programming.

Project 1 - Deep Reinforcement Learning for Peg Solitaire

Purposes:

• Gain hands-on familiarity with Reinforcement Learning (RL) and, in particular, the actor-critic model of RL.
• Learn to use one of the popular deep-learning systems (Tensorflow or PyTorch) and how to integrate it into an RL system as a function approximator

Project 2 - On-Policy Monte Carlo Tree Search for Game Playing

Purposes:

• Implement a general-purpose Monte Carlo Tree Search (MCTS) system for use in a complex 2-person game.
• Learn to employ a neural network as the target policy (and behavior/default policy) for on-policy MCTS.
• Gain proficiency at training policy networks with MCTS and then re-deploying them in head-to-head competitions with other networks.