Implementation of the Q-Learning algorithm.
- QL - algorithm
- DeepQ Learning
- more examples
- SARSA algorithm
Start a server an run index.html
.
Think of following grid world / environment:
Here, R is the Agent in the environment. Its goal is to reach the terminal state G by moving from one field to another. The agent must not go into the traps denoted by '---'.
- 0: NORTH
- 1: SOUTH
- 2: WEST
- 3: EAST
- +1000 at G
- -1000 at trap (---)
- -1 any other field
- -10 if he moves out of the world
The agent is able to solve the problem with the minimum number of moves.