BayesianParameterSampling Class

Reference

Defines Bayesian sampling over a hyperparameter search space.

Bayesian sampling tries to intelligently pick the next sample of hyperparameters, based on how the previous samples performed, such that the new sample improves the reported primary metric.

Initialize BayesianParameterSampling.

Inheritance: azureml.train.hyperdrive.sampling.HyperParameterSampling

BayesianParameterSampling

Constructor

BayesianParameterSampling(parameter_space, properties=None)

Parameters

Name	Description
parameter_space Required	dict A dictionary containing each parameter and its distribution. The dictionary key is the name of the parameter. Note that only choice, quniform, and uniform are supported for Bayesian optimization.
parameter_space Required	dict A dictionary containing each parameter and its distribution. The dictionary key is the name of the parameter. Note that only choice, quniform, and uniform are supported for Bayesian optimization.
properties	Default value: None

Remarks

Note that when using Bayesian sampling, the number of concurrent runs has an impact on the effectiveness of the tuning process. Typically, a smaller number of concurrent runs leads to better sampling convergence. That is because some runs start without fully benefiting from runs that are still running.

Note

Bayesian sampling does not support early termination policies. When using Bayesian parameter sampling, use NoTerminationPolicy, set early termination policy to None, or leave off the early_termination_policy parameter.

For more information about using BayesianParameter sampling, see the tutorial Tune hyperparameters for your model.

Attributes

SAMPLING_NAME

SAMPLING_NAME = 'BayesianOptimization'

Share via