Posted On: Feb 27, 2023
Amazon SageMaker Autopilot, a low-code machine learning (ML) service which automatically builds, trains and tunes the best ML models based on your data, now supports selection of underlying training algorithms while creating an Autopilot experiment. The ability to select algorithms provides you the flexibility to customize your AutoML journey and complete experiments much faster.
Amazon SageMakerAutopilot supports auto or manual selection of two training methods - Ensemble and Hyperparameter Optimization (HPO), to address different machine learning problems. Ensemble and HPO training modes support eight and three algorithms respectively. Each training mode runs a pre-defined set of algorithms on your dataset to train model candidates. By default, Autopilot pre-selects all the available algorithms for the given training mode. Starting today, you can select the algorithm(s) from the list of offered algorithms and customize the Autopilot experiment to meet your model training requirements. Selecting algorithms not only eliminates the need to iterate over non-preferred algorithms but also improve the overall job runtime.
Algorithm Selection in SageMaker Autopilot is now available in all regions where SageMaker Autopilot is available. To get started, Create an SageMaker Autopilot experiment in SageMaker Studio console. You can refer to createAutoMLJob API reference guide for updates to API, and upgrade to the latest version of SageMaker Studio to use the new algorithm selection feature. For more information on this feature, see the developer guide and to learn more about SageMaker Autopilot, visit the product page.