Which search strategy for hyperparameter tuning samples from a distribution of possible values rather than checking every point on a fixed grid?
Show answer
Random search
Random search is often more efficient than grid search because it explores the hyperparameter space more effectively. It is particularly better when some hyperparameters are significantly more influential than others.
Also asked as:
- In hyperparameter tuning, which method involves picking values from a distribution rather than testing every point on a grid?