GitHub - Razbolt/Reinforcement-Learning-in-Chess: Game established with Reinforcement learning to improve itself.

Razbolt / Reinforcement-Learning-in-Chess Public

Notifications You must be signed in to change notification settings
Fork 0
Star 0

Game established with Reinforcement learning to improve itself.

Notifications

Name		Name	Last commit message	Last commit date
Latest commit History 3 Commits
__pycache__		__pycache__
processed		processed
.DS_Store		.DS_Store
.gitignore		.gitignore
README		README
board.py		board.py
generate_training_set.py		generate_training_set.py
train.py		train.py

Repository files navigation

 Zero Knowledge Chest Engine

- Establish the search tree 
* Use a neural net to prune the search tree 

Definition: Value network 

V- f(board)

State of V:

V = 1 White wins
V = 0 Draw
V = -1 Black wins

State(Board):

Pieces(2+7*2 -16):

*Universal
**Blank
**Blank (En passant)  
*Pieces

**Pawn
**Bishop
**Knight
**Rook
**Rook(can castle)
**Queen
**King



Extra state:
* To move 
*Castle available x4   

8x8x4 -4 = 260 bits
8x8x4 - 4 = 260 bits