Exercise 1.5 from Machine Learning by Tom M. Mitchell

• Language used: Java
• Main class: src/Game.java

Unresolved Issues

1. After the first game, the weights are adjusted to give a win condition, but subsequent weight adjustments consistently produce a draw condition
2. Ideal move at each step is always symmetrical (i.e. a positive feature has the same value as another negative feature)
   • causes problems when same value is assigned to all weights for DumbPlayer, as error at each step will always be 0
   • perhaps can consider other features for target expressions

References

1. Machine Learning, Tom M. Mitchell
2. Teaching a Computer to Play TicTacToe, Christopher J. MacLellan (http://www.christopia.net/media/redactor/TicTacToe.py)