next up previous
Next: References Up: Better Trained Ants Previous: Results



By making a modest change to the Santa Fe trail problem we have made it better at training GP. In terms of the Effort required it is approximately three times easier and GP performs about as well as the best results previously obtained using variable length hill climbers [Langdon1998] and size restricted EP search [Chellapilla1997]. By enforcing an optimal size limit on the programs being evolved we are able to do marginally better. Surprisingly GP performance is only marginally affected by a size limit and is roughly constant for wide ranges in maximium size.

This indicates GP is still not greatly out performing random search.

William B Langdon
Wed Mar 4 13:21:44 GMT 1998