EE290S/CS194: Machine Learning for Sequential Decision Making Under Uncertainty

Instructors

 

Anant Sahai

 

Vidya Muthukumar

Schedule

Lecture: 2-3:30 pm Tuesdays and Thursdays

Office hours: immediately after lecture, 258 Cory.

Course communication

Piazza

Prerequisites:

EE126

EE127-227A or CS189-289A.

Grading policy (not curved)

40%: project

20%: midterm (take-home)

15%: team HW (roughly every 2 weeks)

25%: participation

About this course

This course covers fundamental advances in the theory of sequential decision-making from the point of view of statistics, computer science and control. Sequential decision-making is all about learning to make decisions that are embedded in time and in an uncertain environment. What does it even mean to do well in such settings and how can we evaluate performance? What if we do not fully trust a probabilistic model? What if there are game-theoretic or adversarial aspects? How can we intelligently navigate the tension between exploration (figuring out what is going on), exploitation (reaping the rewards of what we have learned), and defense (preventing a potentially adversarial environment from exploiting us!)? How does this change if our feedback from the environment is delayed or sparse?