Modular Multiobjective (Hyper) Neuro-Evolution of Augmenting Topologies + MAP-Elites: Java code for evolving intelligent agents in Ms. Pac-Man, Tetris, and more, as well as code for Procedural Content Generation in Mario, Zelda, Minecraft, and more!
We have the per-objective other score QD plots for MAP Elites that are meant to be used with weighted sum fitness functions. However, for a score with a negative range, you need to subtract out the min score (or rather, just always do this). This affects QD calculation, but nothing else.
HOWEVER: I now wonder ... do regular MAP Elites fitness scores account for this kind of adjustment? What if the standard fitness can be negative? May need to fix that too.
schrum2
commented
1 year ago
We have the per-objective other score QD plots for MAP Elites that are meant to be used with weighted sum fitness functions. However, for a score with a negative range, you need to subtract out the min score (or rather, just always do this). This affects QD calculation, but nothing else.
HOWEVER: I now wonder ... do regular MAP Elites fitness scores account for this kind of adjustment? What if the standard fitness can be negative? May need to fix that too.