Learning Game Playing

Learning Game Playing

Learning
vs. People Learning
by Self-Play

Samuels' Checkers

Tesauro's TD-Gammon
PROBLEMS
* Too many games to be played
* Noisy human performance
* Human learning * Lack of generalization
* Deceptive landscapes
* Convergence

Learning vs. People	Learning by Self-Play
Samuels' Checkers	Tesauro's TD-Gammon
PROBLEMS
* Too many games to be played * Noisy human performance * Human learning	* Lack of generalization * Deceptive landscapes * Convergence