Evolutionary Computation Fitness Landscapes

The goals of this project are:

1. Catalog a variety of different fitness landscapes used in evolutionary computation
2. Catalog a variety of different properties that people measure about fitness landscapes
3. Measure those propreties on those fitness landscapes
4. Measure which evolutionary algorithms are best on which landscapes
5. Look for correlations between the results of steps 3 and 4.

Some fitness landscapes to check out

• NK Landscapes (and variants)
• NM Landscapes
• Function optimization problems (GECCO Niching Competition)
• Program Synthesis Benchmark Suite
• Input-output machine learning (Dow Chemical, this, others)