0

我有一个内部开发的应用程序,并希望将 ML.Net 机器学习库与自定义分类器结合起来。特别是,如何制作与LearningPipeline该类一起使用的自定义分类器?这样做时如何制作自定义分类器意味着遵守尽管是密封类 但仍ILearningPipelineItem需要Microsoft.ML.Runtime.EntryPoints.Var作为属性的接口?Microsoft.ML.Runtime.EntryPoints.Var

4

1 回答 1

0

LearningPipeline is not meant to be extensible. There is no easy way to add a custom classifier.

You could take a look at the new API (start here), and implement an IEstimator<ITransformer>, which you will then be able to plug into the pipeline and Fit.

It is going to be possible, but still not exactly easy.

于 2018-10-04T04:19:43.410 回答