AWS Clean Rooms supports incremental and distributed training for custom modeling

AWS Clean Rooms now supports two enhancements to its machine learning capabilities that help you train models more efficiently and at scale to generate predictive insights in a Clean Rooms collaboration. Incremental training enables you to build upon existing model artifacts to create new models, and distributed training allows you to train models across multiple compute instances simultaneously. These capabilities help data scientists and ML practitioners accelerate data collaboration and analysis while maintaining the privacy of the training datasets.

With AWS Clean Rooms ML custom modeling, you and your partners can train and run inference on a custom ML model using collective datasets at scale without having to share sensitive intellectual property. With incremental training, you can leverage previously trained models to create new variants using expanded datasets, significantly reducing training time and compute resources. Additionally, distributed training lets you process large-scale datasets efficiently by distributing the training workload across multiple instances.

AWS Clean Rooms ML helps you and your partners apply privacy-enhancing controls to safeguard your proprietary data and ML models while generating predictive insights—all without sharing or copying one another’s raw data or models. For more information about the AWS Regions where AWS Clean Rooms ML is available, see the AWS Regions table. To learn more, visit AWS Clean Rooms ML.