Ghaith81
commented
1 year ago Hi Tobi, Thank you for reaching out. The approach in the paper is independent from the metaheuristic, you could basically use any variant of GA. I'm also including an implementation of PSO, a completely different family of optimizers and observing similar results. NSGA2 was not my first choice because in my work I have considered single objective optimization. But it could be great to run a multi objective implementation with NSGA2 instead of CHC. Having said that, I was presenting this work in GECCO23, the largest conference of evolutionary computation, during one of the keynote sessions it was mentioned that CHC is considered a trade secret algorithm, which explain why you wouldn't find a lot of good implantations, other than mine ;) Let me know if you need any support or you face difficulties running the repo. Cheers! Ghaith
From: TobiArndt @.> Sent: Wednesday, July 12, 2023 2:07 PM To: Ghaith81/Fast-Genetic-Algorithm-For-Feature-Selection @.> Cc: Subscribed @.***> Subject: [Ghaith81/Fast-Genetic-Algorithm-For-Feature-Selection] NSGA2 (Issue #1)
Hey, thanks for your great work and for publishing the paper related code base! I am really interested in the fact that you didn't use NSGA2 algorithm. Have you considered it or are there special reasons? Or do you think that it could work for big datasets as well? I am quiet new in field of genetic algorithms so I would be happy hear your oppinion to the topic of using NSGA2 for your paper instead of CHC.
— Reply to this email directly, view it on GitHubhttps://github.com/Ghaith81/Fast-Genetic-Algorithm-For-Feature-Selection/issues/1, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AJCIL5O6VWWWMW2QNFLH3EDXP2OQZANCNFSM6AAAAAA2HO3JCQ. You are receiving this because you are subscribed to this thread.Message ID: @.***>
Hey, thanks for your great work and for publishing the paper related code base! I am really interested in the fact that you didn't use NSGA2 algorithm. Have you considered it or are there special reasons? Or do you think that it could work for big datasets as well? I am quiet new in field of genetic algorithms so I would be happy hear your oppinion to the topic of using NSGA2 for your paper instead of CHC.