Skip to content
/ KE_AE Public

Reinforcement Learning- Watkins Q Learning(Eligibility Traces) for Maze Solving Agent

Notifications You must be signed in to change notification settings

MuaazBin/KE_AE

Repository files navigation

KE_AE

Teknomo's Q Learning Tutorial MATLAB Code is adapted for our coursework. Graphs are added and Eligibility Traces is incorporated

File 1 ReinformcentLearningGreedy.m

This file has normal Q Learning with Epsilon Greedy Policy The csv file being read in matlab code has to be changed depending on Case (Reward Matrix) The values of constant Epsilon , Gamma, Alpha to be changed as per Case and which states to be read if punishment incorpoated in Case 5A and 5B in code line 49

File 2 ReinforcementLearningQlambdaWatkins.m

It is Watkin's Q Learning Algorithm implementation

Reward Matrixes in CSV Files RewardMatrix25 was just for testing and can be ignored

By: Muaaz Bin Sarfaraz and Chadi El-Hajj These files should be refered with the actual coursework report.

email: muaazbinsarfaraz@yahoo.com

Ref:http://people.revoledu.com/kardi/tutorial/ReinforcementLearning/

MAZE to be solved by agent graphmaze1

About

Reinforcement Learning- Watkins Q Learning(Eligibility Traces) for Maze Solving Agent

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages