Skip to content

nimchimpski/Othello-Q-learning

Repository files navigation

Othello Player. Reinforcement learning with q-Table

CLICK TO PLAY!

In Train function:

A model playes itself, (always as black using state inversion). Epsilon Greedy with a decay function ensures sufficient exploration outside the previously learnt policy. A Heuristic evaluation function is used after every move to update the q-table.

In Evaluate function:

The model, using a specified q-table, playes against a random player.
Win rate and win/loss ratio are recorded.

Flask

A flask app presents a web page game where the user can play a model, using a predetermined Q-table. The baord is updated with AJAX.

About

Reinforcement Learning game engine for Othello

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published