Amazon’s SageMaker Autopilot Now Automates AWS Machine Learning Models. Amazon has announced SageMaker Autopilot, a new tool that Amazon claims can help boost machine learning projects through automating tasks such data preprocessing, training parameters, and classification. SageMaker, the company’s managed service to machine learning, allows companies to work in the Studio IDE environment to “stitch” together different tools such as: – Architectural Scale Models

SageMaker Neo, a learning tool
SageMaker AugmentedAI, for “human review” of model predictions
SageMaker Model Tuning is an automated optimization
SageMaker Autopilot is now available for automating the building and training of machine learning models.

Amazon shared the following information about SageMaker Autopilot: SageMaker Autopilot first examines your data set and then runs a number candidates to determine the optimal combination data preprocessing, machine learning algorithms, and hyperparameters. This combination is then used to train an Inference Pipeline that you can deploy on a real time endpoint or batch processing. All of this happens on fully-managed infrastructure, as is the case with Amazon SageMaker. SageMaker Autopilot generates Python code that explains how the data was preprocessed. This allows you to understand what SageMaker Autopilot did and you can reuse that code for manual tuning if needed. Amazon emphasizes that SageMakerAutopilot allows you to inspect what’s underneath, unlike other tools or “black box” models. Further down the page, you will find a detailed tutorial on how to use the Autopilot tool. SageMaker Autopilot is now available and can be accessed via an Amazon SageMaker subscription. You can find it here.