ApproximatedKernelMappingEstimator 类

定义

将矢量列映射到低维特征空间。

public sealed class ApproximatedKernelMappingEstimator : Microsoft.ML.IEstimator<Microsoft.ML.Transforms.ApproximatedKernelTransformer>
type ApproximatedKernelMappingEstimator = class
    interface IEstimator<ApproximatedKernelTransformer>
Public NotInheritable Class ApproximatedKernelMappingEstimator
Implements IEstimator(Of ApproximatedKernelTransformer)
继承
ApproximatedKernelMappingEstimator
实现

注解

估算器特征

此估算器是否需要查看数据来训练其参数?
输入列数据类型 已知大小的矢量 Single
输出列数据类型 已知大小的矢量 Single
可导出到 ONNX

生成的 ApproximatedKernelTransformer 结果创建一个新列,该列在输出列名称参数中指定,其中每个输入向量映射到特征空间,其中内部产品大约是两个移位固定内核函数之一:Gaussian 内核或 Laplacian 内核之一。 通过将特征映射到近似非线性内核的空间,线性方法可用于近似更复杂的内核 SVM 模型。 此映射基于 Rahimi 和 Recht Large-Scale 内核计算机的论文随机特征

有关使用示例的链接,请查看“另请参阅”部分。

方法

Fit(IDataView)

火车并返回一个 ApproximatedKernelTransformer

GetOutputSchema(SchemaShape)

返回 SchemaShape 由转换器生成的架构。 用于管道中的架构传播和验证。

扩展方法

AppendCacheCheckpoint<TTrans>(IEstimator<TTrans>, IHostEnvironment)

将“缓存检查点”追加到估算器链。 这将确保针对缓存的数据训练下游估算器。 在执行多个数据传递的训练器之前,拥有缓存检查点会很有帮助。

WithOnFitDelegate<TTransformer>(IEstimator<TTransformer>, Action<TTransformer>)

给定估算器后,返回将调用委托的 Fit(IDataView) 包装对象。 估算器通常必须返回有关拟合情况的信息,这就是为什么 Fit(IDataView) 该方法返回特定类型化对象的原因,而不仅仅是常规 ITransformer对象。 但是,同时, IEstimator<TTransformer> 通常形成为包含许多对象的管道,因此,我们可能需要通过 EstimatorChain<TLastTransformer> 估算器链生成一系列估算器,以便我们要获取转换器的估算器被埋在此链中的某个位置。 对于这种情况,我们可以通过此方法附加调用一次将调用的委托。

适用于

另请参阅