https://github.com/kkkaan/q-learning-openai-gym

A simple implementation of Q-learning algorithm.
https://github.com/kkkaan/q-learning-openai-gym

openai openai-gym q-learning reinforcement-learning

Last synced: 6 days ago
JSON representation

A simple implementation of Q-learning algorithm.

• Host: GitHub
• URL: https://github.com/kkkaan/q-learning-openai-gym
• Owner: kkKaan
• Created: 2023-09-04T18:53:16.000Z (about 1 year ago)
• Default Branch: main
• Last Pushed: 2024-02-01T06:27:57.000Z (10 months ago)
• Last Synced: 2024-02-01T08:33:59.745Z (10 months ago)
• Topics: openai, openai-gym, q-learning, reinforcement-learning
• Language: Python
• Homepage:
• Size: 11.7 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Q-Learning Agent for CliffWalking

## Project Overview

This project implements a Q-Learning agent to solve the CliffWalking environment from OpenAI Gym. The agent is trained to navigate a grid world environment, avoiding cliffs and finding the shortest path to the goal.

## Features

- Implementation of the Q-Learning algorithm.

- Epsilon-greedy strategy for action selection.

- Training and testing phases for performance evaluation.

- Ability to save and load trained Q-tables.

## Requirements

- Python 3.x

- OpenAI Gym

- NumPy

## Usage

1. Run the script: `python3 main.py`

2. Follow the prompt to load an existing Q-table or train a new agent.

## Q-Learning Agent

The agent is designed to:

- Learn optimal policies via Q-Learning.

- Use an epsilon-greedy strategy for a balance between exploration and exploitation.

## Training

- The agent is trained over a specified number of episodes, learning to maximize rewards in the CliffWalking environment.

- The Q-table records the value of taking certain actions in specific states.

## Testing

- The agent's performance is evaluated over a number of test episodes.

- Rewards per episode are recorded to gauge the effectiveness of the learned policy.