Introduces verifiable machine learning, focusing on the limitations of current learning-based methods regarding safety and robustness. Explores adversarial robustness, uncertainty quantification, safety verification, reachability analysis, and safe/robust reinforcement learning. Examines neural network relaxations and their applications in robotics, control, image classification, medical diagnosis, and language modeling.
Offering history
Term
Sections
Enrolled
Capacity
Full
Open seats/section
Fall 2025
1
16
23
70%
7.0
Snapshots from scheduled scrapes — not live seat availability. "Full" can exceed 100% when sections over-enroll.
Meeting times
Share of recent sections by weekday: M 100% · T 0% · W 100% · Th 0% · F 0%
Common patterns: MW (100% of sections)
— in patterns, R means Thursday