Trustworthy deep learning

Stanley Osher
University of California, Los Angeles (UCLA)

In this work we discuss trustworthy deep learning, which includes
1. Robust deep learning, 2. Accurate deep learning, 3. Efficient deep learning and 4. Private deep learning,
with theoretical guarantees, usually based on ideas borrowed from differential equations.

(joint work with Bao Wang and others)

Presentation (PDF File)

Back to Deep Learning and Medical Applications