|
Click on a filename to view its contents
|
Game specific functions (MatLab m-files)
|
interface.m |
Main interface to the NN-Seven card game training functions.
|
playgame.m |
Plays one game of Seven, updating the neural network weights.
|
playgame_rand.m |
Plays one game of Seven with two random players and the AI.
|
playcard.m |
The player makes a move.
|
playcard_rand.m |
The player makes a move (possibly random).
|
givecard.m |
The player passes his/her worst card.
|
givecard_rand.m |
The player passes his/her worst card (possibly random).
|
findmoves.m |
Determines all moves that can be made from the given hand.
|
makeplay.m |
Play the given card and update the current game state.
|
iswinner.m |
Check the end condition of the game.
|
|
Neural Network specific functions (MatLab m-files)
|
encode.m |
Encodes the current game state as an input vector for the network.
|
evaluate.m |
Implements a standard forward propagation.
|
train.m |
Trains the neural network using a modified version of temporal diff.
|
learn.m |
Implements standard back propagation.
|
sigma.m |
Implements the sigmoid function: y = 1 / (1 + exp(-x)).
|
|
Game Interface (JavaScript)
|
nn-seven.zip |
JavaScript web interface for the NN-Seven card game (zipped file).
|