Skip to content

Q-learning agent to solve the frozen lake problem from the OpenAI gym

Notifications You must be signed in to change notification settings

pagrim/FrozenLake

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

47 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Frozen Lake

The code in this repository aims to solve the Frozen Lake problem, one of the problems in AI gym, using Q-learning and SARSA

Algorithms

The FrozenQLearner.py file contains a base FrozenLearner class and two subclasses FrozenQLearner and FrozenSarsaLearner. These are called by the experiments.py file.

Experiments

The experiments.py file contains the details of the experiments run using the two algorithms. If the output_file parameter of the methods is set to true a CSV file summarising each session will be written to the outputs directory. The experiments call the FrozenLearner subclasses.

Analysis

The analysis.R file was used to analyse the output of experiments using charts and some quantitative analysis. The charts are saved to the plots directory in some cases

Requirements to Run

A python 3.x environment with gym, numpy, sklearn (tested on python 3.5) to run the experiments. An R environment with ggplot2, dplyr, TTR, reshape2, stringr to run the analysis.

Example outputs

The total number of steps and number of random steps

random steps in each episode

Whether each episode resulted in reaching the goal 'G' or a hole 'H'

moving average steps each episode

The total reward for each episode

increasing reward each episode

About

Q-learning agent to solve the frozen lake problem from the OpenAI gym

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published