Domain adaptation, transfer learning, multitask learning, federated learning, meta learning, representation learning, few-shots learning, lifelong learning, robust optimization, the list goes on: these are all important recent directions in machine learning that are concerned with learning in heterogeneous and ever-changing environments, as motivated by modern applications.

While these areas are often studied separately, they naturally share many central questions: for instance, what information a data distribution may have about another, and how to leverage such information to speed up learning across related environments. Our understanding of these problems is still fledgling, and design decisions remain ad-hoc despite many successes observed in practice. 

The workshop aims to bring together theoretical and applied researchers at the forefront of these areas, from both academia and industry, to not only present their latest findings, but to also identify common threads and foster new collaboration on important future directions.

If you require special accommodation, please contact our access coordinator at simonsevents [at] berkeley.edu with as much advance notice as possible.