Pareto front solutions are too many


I have a moo problem with 12 fitness objectives. I did my optimization with 50 generation size and 100 generation count. I got 1148 pareto front solutions and last generations’s all solutions are in pareto front. I have no idea if there is a bias or problem, or this is normal.

I know that 12 is too much but and in this complexity I probably get too much pareto results. But still, is this a problem?


Well … assuming that you cut the MOO mustard solely via code … the pareto vis approach has little meaning for more than 2 objectives (unless you can imagine things in a multi dimensional Universe). Try to get rid of the pareto and mastermind other ways to visualize results (if that is what you need - but why?). See visual results for 3 objectives (could be 666, mind).

BTW: In MOO the biggest thing is the normalization of values.

A very short answer is that 12 fitness objectives might be too many. My recommendation is to reformulate your probelm differently to have less objectives