/B659-RL-Project

Primary LanguageJupyter Notebook

CSCI B659 - Reinforcement Learning Final Project

Developers: Saber Sheybani, Wesley Liao

Spring 2020

Notes for running the project:

The .py scripts support the Jupyter Notebook files. The report was generated using the expx.y.py files.

  • TabularDyad: figure 2. Dyad of RL agents employing tabular SARSA to solve the Dyadic Slider game.
  • DQNDyad: figure 3-5. Dyad of RL agents employing DQN to solve the Dyadic Slider game. Experiments include exploring the performance and the convergence of the multiagent system in various hyperparameter settings and task conditions.