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.