You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
For paper doing CMA-ES on hyperparameters, we might want to compare with "Bayesian optimisation", which has been called the gold standard for tuning [citation needed].
As was the case CMA-ES, for which Arnaud did a "plain" implementation, it might be necessary to implement one of these methods ourselves, so that we're testing the core idea behind the method, not some battle-hardened competition code.
Then again, maybe we want to save ourselves the trouble and use an industry-standard method?
For paper doing CMA-ES on hyperparameters, we might want to compare with "Bayesian optimisation", which has been called the gold standard for tuning [citation needed].
As was the case CMA-ES, for which Arnaud did a "plain" implementation, it might be necessary to implement one of these methods ourselves, so that we're testing the core idea behind the method, not some battle-hardened competition code.
Then again, maybe we want to save ourselves the trouble and use an industry-standard method?
See also #684
The text was updated successfully, but these errors were encountered: