Randomized Grid Search is a variation of Grid Search that samples each parameter from a distribution. Conventional grid search evaluates the model at fixed combinations of parameter values and could be very inefficient as the complexity grows exponentially with the number of parameters. Randomized Grid search is a more efficient and an effective alternative.
How to tune hyperparameters with Randomized Grid Search?
Posted on