https://github.com/omaraflak/reinforcement-learning-cpp

Reinforcement Learning algorithm from scratch in C++.
https://github.com/omaraflak/reinforcement-learning-cpp

cpp from-scratch neural-network qlearning reinforcement-learning

Last synced: 2 months ago
JSON representation

Reinforcement Learning algorithm from scratch in C++.

• Host: GitHub
• URL: https://github.com/omaraflak/reinforcement-learning-cpp
• Owner: omaraflak
• Created: 2018-07-19T11:18:53.000Z (over 6 years ago)
• Default Branch: master
• Last Pushed: 2018-07-19T11:29:00.000Z (over 6 years ago)
• Last Synced: 2024-03-18T00:35:43.616Z (10 months ago)
• Topics: cpp, from-scratch, neural-network, qlearning, reinforcement-learning
• Language: C++
• Size: 6.84 KB
• Stars: 8
• Watchers: 3
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Reinforcement Learning

This code demonstrates how to make a reinforcement learning algorithm from scratch in C++.

# Download, Compile & Run

```

git clone https://github.com/OmarAflak/Reinforcement-Learning-CPP

cd Reinforcement-Learning-CPP

make

./main

```

# Environment

The agent has to go from point A to point B.

    A..........B

    

He has two available commands : `left` and `righ`.

The reward policy is the following :

* if (agent reaches point **B**) { reward=1 }

* else { reward=0 }

# Network

Although the problem is very simple, I used a Neural Network to approach the **Q(s,a)** function.

The network takes the current state of the game as an input, and it outputs the **Q-value** for each possible action.