Tabular Learning: Reinforcement Learning Algorithms

Overview

This repository contains implementations of various tabular reinforcement learning algorithms for solving sequential decision problems. The focus is on value-based methods, where the State-Action function (Q(s, a)) is estimated and maintained in a tabular format, commonly known as a Q-table. These algorithms are tested on grid-based environments to compare their performance under different settings and exploration strategies.

Features

• Dynamic Programming: The agent has access to all properties of the environment
• Model-Free Algorithms:
  • SARSA: On-policy temporal difference learning.
  • Q-Learning: Off-policy temporal difference learning.
  • Monte-Carlo Methods: Learning from complete episodes.
• Exploration Strategies:
  • ϵ-greedy
  • Boltzmann exploration
  • Upper Confidence Bound (UCB)

Prerequisites

Ensure you have Python 3.x installed along with the necessary libraries:

pip install -r requirements.txt

Installation

1. Clone the repository:

   git clone https://github.com/Dinu23/Tabular-Learning.git

2. Navigate to the project directory:

   cd Tabular-Learning

Running the Algorithms

You can run the algorithms directly through the provided scripts

Example command to run Q-Learning:

python Q_learning.py

Contact

For any questions or inquiries, please reach out to:

• Name: Dinu Catalin-Viorel
• Email: viorel.dinu00@gmail.com