search

BIMK / PlatEMO

Evolutionary multi-objective optimization platform
1.58k stars 463 forks source link

Hyper parameter adjust through platemo function #134

Open war3water opened 1 year ago

war3water commented 1 year ago

Hello！

This is a impressive framework for MOP problems solving! If I not misunderstanding about framework of PlatEMO, Here are some tips from my side:

  1. When I do research to compare different algorithm performance, I found here is little hyperparameter can be adjust for most algorithm, like hyperparameter of crossover\mutation etc... but these hyperparameter will affect final performance of algorithm a lot. I know use self definition framework of PlatEMO will solve this problem, but it is more like rewrite the algorithm or change source code not only fine tuning some parameter to see how it will change. So could PlatEMO can support we change hyperparameter in 'platemo' function? Even in GUI we could see NSGA-II not have the directly hyperparameter adjust item like 'probility of crossover and mutation'.
  2. Based on 1, could we make PlatEMO input framework hyperparameter easy to change in main function, like pymoo or Geat2py? Thanks for your team great works!