Unsupervised Machine Learning using a Neural Network

I made this AI focuses on ‘learning’ from playing many games of chess over time. Initially, the AI had a neural network with randomly weighted nodes. As it played, it would parse moves through the networks to determine which ones it thinks are best. Once the other player makes a move, it will backward propagate the result of their move through the network to adjust the weights.

This AI was not fully finished as it needed many more games to generate a smarter neural network.

Maximin/Minimax using Alpha-Beta Pruning

I made this AI implemented the minimax algorithm which made use of Alpha-Beta pruning and Zero-sum pruning. This makes this AI a depth-first sorting algorithm.

With the Maximin algorithm, I got the AI to minimize the worst possible scenario that could occur from the next player's move. This is done by looking at all possible moves and all possible moves the other player could do from there, this determined the worst possible combinations of moves and whichever returned to best value was picked.

This process can take a long time, especially if the AI is looking several moves ahead, to minimize this time, I used a mix of Alpha-Beta Pruning and Zero-sum Pruning. I implemented Alpha-Beta Pruning to cut off nonviable paths by ending a search completely when at least one possibility has been found that proves the action to be worse than a previously examined action. Zero-sum Pruning allowed my to use the same code inversely when checking the opponent's possible moves

Optimal Decision Making with Uncertainty

I made this AI focused on the idea of Optimal Decision Making. Through the use of many variables, I had it figure out the optimal move to take next.

Optimal Decision Making has three main parts:

1. Predicting the outcome of a decision

2. Assigns a value to that decision

3. Finds the decision that maximizes the value

This allowed the AI to makes judgements of each possible move then do the one that gives itself the greatest advantage over the opponent. This AI had the highest win-rate when play against the other AIs.

Chess Base