Skip to content

Latest commit

 

History

History
18 lines (12 loc) · 685 Bytes

File metadata and controls

18 lines (12 loc) · 685 Bytes

Reinforcement Learning Based on Monte Carlo Method (gym API FrozenLake8x8-v0)

This program is to solve the FrozenLake8x8 with the MC control method.

In this program, On-policy First Visit Monte Carlo Control was implemented on both none-slippery and slippery FrozenLake8x8.

major parameters:
n_episodes = 500000
current_epsilon = 1.0
max_epsilon = 1.0
min_epsilon = 0.001
decay_rate = 0.0001

After 500000 episodes, the results show that none-slippery FrozenLake8x8 scored 0.9794 and slippery FrozenLake8x8 scored 0.3113.

File was compiled on Anaconda Jupyter Notebook and Python 3.8 environment.

https://gym.openai.com/envs/FrozenLake-v0/