ease.ml – integrating AutoML into SDSC Eco-Systems

Objectives:

• Integrate ease.ml into the SDSC eco-system (in particular, the RENKU platform) and provide model selection functionalities to a subset—mainly deep-learning based applications over images, text, and time series—of SDSC applications;
• Work closely with the SDSC to further advance the state of the art of AutoML, including overcoming a number of computer science challenges that mainly involve developing novel techniques for scalable Bayesian optimization;
• Pursue the emerging direction of how to automatically manage ML pipelines in a data-driven fashion.