James Bergstra & Yoshua Bengio (2012)
Journal of Machine Learning Research, 13, 281-305.
URL: https://jmlr.org/papers/v13/bergstra12a.html
Abstract. Shows that random search is substantially more efficient than grid search for hyperparameter tuning when only a few dimensions matter, since random sampling explores the important directions more thoroughly for the same computational budget.
Tags: hyperparameters random-search url-only
Cited in: