online papers

Here's every online paper I know of which isn't mentioned elsewhere on the web site. If yours should be here, let me know about it.
Jay : game learning : papers
papers added in January
papers added in September
-->
Evolutionary induction of subroutines (1992, 6 pages)Peter J. Angeline and Jordan PollackCompressed postscript. A genetic programming method which includes learning of subroutines is tested on a tic-tac-toe task.
Coevolving high-level representations (1993, 15 pages)Peter J. Angeline and Jordan PollackCompressed postscript. An expanded version of the above paper, "Evolutionary induction of subroutines," with new results. Here, the subroutines are described as "modules".
Competitive environments evolve better solutions for complex tasks (1993, 8 pages)Peter J. Angeline and Jordan PollackCompressed postscript. Advocates competitive learning for genetic algorithms. In an experiment, tic-tac-toe players evolved by genetic programming were more robust (though not stronger) when a fitness function was defined by tournament play rather than by play against a fixed opponent.
Adaptive critic design in learning to play game of go (1997)Raonak Zaman, Danil Prokhorov, and Donald C. Wunsch IIWeb page. The authors trained a neural network, using what they call a Heuristic Dynamic Programming (HDP) adaptive critic. After 6000 training games, the network had learned to defeat Wally, a weak free program, 1% of the time.
updated 3 January 2001