HyperParameterSampling 类

所有超参数采样算法的抽象基类。

此类封装了派生采样类的超参数空间、采样方法和其他属性:BayesianParameterSamplingGridParameterSamplingRandomParameterSampling

继承
HyperParameterSampling

构造函数

HyperParameterSampling(sampling_method_name, parameter_space, properties=None, supported_distributions=None, distributions_validators=None)

参数

sampling_method_name
str
必需

采样方法的名称。

parameter_space
dict
必需

包含每个参数及其分布的字典。

properties
dict
默认值: None

带有算法附加属性的字典。

supported_distributions
set[str]
默认值: None

受支持的分发方法的列表。 默认值为“None”表示支持所有分发,如模块 parameter_expressions 中所述。

distributions_validators
默认值: None

方法

to_json

返回表示超参数采样对象的 JSON。

to_json

返回表示超参数采样对象的 JSON。

to_json()

返回

JSON 格式的采样策略。

返回类型

str