By using AWS re:Post, you agree to the Terms of Use

Hyperparameter tuning for pipeline model


We are developing a machine learning algorithm that creates additional features. These features should then be fed into eg a classification algorithm (SVM, XGboost,...). We are using a pipeline to put the two units together, and now want to do hyperparameter tuning for the whole pipeline. Basically what we want to do is:

model = PipelineModel(..., models = [ our_model, xgb_model ])
deploy = Estimator(image_uri = model, ...)
tuner = HyperparameterTuner(deply, .... tune_parameters, ....)

but there are always errors somewhere along the line.

Is there any documentation or example how to achieve something like the above?


2 Answers

Hello Norbert,

I believe you are referring here the SageMaker pipelines and how that can be incorporated with the Hyper parameter turning. (Please correct me otherwise.)

Following notebook provides and example of how a Hyperparameter Tuning Job can be run as a step in a SageMaker Pipeline.

answered 7 months ago
  • Hi thanks for your comment and the link! Yes, we want to combine our own feature extraction algorithm (container based) with a standard classification system, plug them together via a sagemaker pipeline, and the do hyperparameter tuning across the whole.

    We will look into the provided sagemaker notebook.

    Thanks again




Again, thanks for your answer, but after study it turns out that you are speaking about "Sagemaker Workflow Pipelines", while I need hyper-parameter tuning for models created via sagemaker.pipeline.ModelPipeline

Do you have any suggestion concerning hyperparameter tuning of modelpipelines?

answered 7 months ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions