Open deil87 opened 6 years ago
We should somehow share(not maybe physically) hyperparameters instances for all classifiers within one ensemble individual/generation/evolution. How to find similarities between classifiers in terms of optimal hyperparameters settings?
Some parameter could be explored independently from others? Some depends on other and we should take into account all the combinations :(
We should somehow share(not maybe physically) hyperparameters instances for all classifiers within one ensemble individual/generation/evolution. How to find similarities between classifiers in terms of optimal hyperparameters settings?