K-Means, GMM, and Reinforcement Learning

• Unsupervised Learning-Clustering: Experimented and evaluated both k-Means and EM algorithm with Gaussian mixtures on different data clusters.

• Reinforcement Learning: Implemented Qlearning algorithms with both epsilon-greedy and softmax policy in a stochastic maze environment.

Implementations include:

• maze.py defines the MazeEnv class, the simulation environment which the Q-learning agent will interact in.
• qlearning.py defines the implemented qlearn function, along with several helper functions.
• plotting_utils.py: defines several plotting and visualization utilities.
• K-Means-GMM-and-Reinforcement-Learning.ipynb the final jupyter notebook project.