Archived Repository The code of this repository was written during a Marmelab agile integration. It illustrates the efforts of a new hiree, who had to implement a board game in several languages and platforms as part of his initial learning. Some of these efforts end up in failure, but failure is part of our learning process, so the code remains publicly visible. This code is not intended to be used in production, and is not maintained.

connect-four-tensorflow

Connect four solver using neural networks and tensorflow

Run

Runs the script

python train.py --p1 <xxx> --p2 <xxx> --board_width <x> --board_height <x> --iterations <x> --randomness <x> -v

• p1 : which algorithm the first player should use (human, minimax, neural_network)
• p2 : which algorithm the second player should use (human, minimax, neural_network)
• board_width : width of the board (default: 4)
• board_height : height of the board (default: 4)
• iterations : number of games played (default: 1)
• randomness : randomness of first player (default: 0.25)
• v: verbose output

Example :

python train.py --p1 minimax --p2 neural_network --iterations 200000 --randomness 0.1

Test

Runs all the tests of the project

make test

Contribute

Where the things happen :

• train.py : handles the differnt games and calls the different AIs
• ai/neural-network/neural-network.py : create the neural network, uses it for prediction and handles back propagation
• connectfour/game.py : handles connect four game, makes AIs play and check who wins