NicMcPhee / XO-bias-study

Results and write-up of our genetic programming crossover bias study
MIT License
0 stars 0 forks source link

Do tournament sizes of 3 and 5 #15

Closed NicMcPhee closed 9 years ago

NicMcPhee commented 9 years ago

It would be interesting to see where the system flattens out.

NicMcPhee commented 9 years ago

It looks like 5 and 7 are very similar (even on hard problems like Ordertree), so I think it was good to add 3, 5, and 7, but I don't think there's any need to go bigger than that.