/hDQN2018

This is a course project for Reinforcement Learning: Theory and Algorithms, School of Mathematical Sciences, Peking University, 2018 spring.

Primary LanguagePython

An Empirically Study of Hierarchical Deep Reinforcement Learning

This is a course project for Reinforcement Learning: Theory and Algorithms, School of Mathematical Sciences, Peking University, 2018 spring.

In this project we implemented two reinforcement learning environment with sparse rewards, Discrete Stochastic Dicision Process and Maze Game.

Run corresponding python files to train DQN, hDQN or hDQNm for tasks.

Results and more details in report.