AutoML

AutoML on ML Workbench is currently a beta feature.

AutoML is a subset of machine learning with the goal of reducing the expertise needed to effectively run machine learning experiments. It assists the user in tuning the hyperparameters for model training.

ML Workbench uses the open-source Katib project through Kubeflow to simplify process of running machine learning experiments.

Screenshot of the Experiments page upon first launch

Tutorial notebook

We have a sample Docker file and notebook for learning to use AutoML (Katib) here: Hyperparameter Tuning with Katib. The notebook is a self-contained tutorial package that guides you through learning to use the functionality.