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
Hi, if there are hundreds of hyperparams, can I use open-box to get a good result? And is there a empirical relationship between run times and number of hyperparams?
The text was updated successfully, but these errors were encountered:
yuzizbk
changed the title
Hi, if there are hundreds of hyperparams, can I use open-box to get a good result? And is there a empirical relationship between run times and number of hyperparams?
How many hyperparams can open box can handle?
Jan 12, 2022
Given a higher-dimensional search space (with more hyperparameter to optimize), hyperparameter tuning tools (e.g., openbox) need more trials to reach a good result.
Please try to use probabilistic random forest as surrogate model (set surrogate_type='prf') to deal with high-dimensional problems (with hundreds of hps). Prf is also faster than other surrogates such as GP. In this paper[1], prf performs best in Auto-WEKA problem with 786 hps.
Other methods to tackle high-dimensional problems are under developing.
Ref:
[1] K. Eggensperger, M. Feurer, F. Hutter, J. Bergstra, J. Snoek, H. Hoos, and K. Leyton-Brown. Towards an empirical foundation for assessing bayesian optimization of hyperparameters. In NIPS workshop on Bayesian Optimization in Theory and Practice, 2013.
Hi, if there are hundreds of hyperparams, can I use open-box to get a good result? And is there a empirical relationship between run times and number of hyperparams?
The text was updated successfully, but these errors were encountered: