ESE Spring Seminar – “End-to-end Learning for Robust Decision Making”

Because the physical world is complex, ambiguous, and unpredictable, autonomous agents must be engineered to exhibit a human-level degree of flexibility and generality — far beyond what we are capable of explicitly programming. Achieving such rich and intricate decision making requires rethinking the foundations of intelligence across all stages of the autonomous learning lifecycle.

In this talk, I will share new learning-based approaches towards dynamic, resilient, and robust decision making of autonomous systems. Such solutions are capable of not only reliably solving a particular problem, but also anticipating what could go wrong in order to strategize, adapt, and continuously learn. We advance robust decision making by (1) computationally designing rich synthetic environments of hard to collect, out-of-distribution edge-cases; (2) creating efficient, expressive, and interpretable learning models; and (3) developing adaptive, robust, and grounded learning algorithms, and exploiting their interdependence to realize generalizable decision making.