Program

Note: The time zone is Pacific Daylight Time (PDT).

Opening Remarks: 9:00-9:05am

SESSION 1: 9:05-10:50am (4 slots)

• 9:05 - 9:30 Andreas Krause: Towards Provably Safe Reinforcement Learning
• 9:30 - 9:55 Nils Jansen: Planning under Uncertainty – AI Safety via Formal Verification
• 9:55 - 10:00 Short break
• 10:00 - 10:25 Iain Whiteside: AI Safety for Automated Driving
• 10:25 - 10:50 Krishnamurthy Dvijotham: Generalized black-box verification via randomized smoothing

Coffee Break: 10:50-11:15am

SESSION 2: 11:15am-12:05pm (2 slots)

• 11:15 - 11:40 Sriram Sankaranarayanan: Learning and Verifying Conformant Data-Driven Models of Dynamical Systems
• 11:40 - 12:05 Dorsa Sadigh: What Happens to Safety When Humans are not Optimal?

Lunch Break: 12:05pm-1:15pm

SESSION 3: 1:15-2:05pm (2 slots)

• 1:15 - 1:40 James Lopez: Ensuring Safe Operation of AI for Unmanned Aircraft Systems
• 1:40 - 2:05 Suman Jana: Training Verifiably Robust Malware Classifiers

Coffee Break: 2:05-2:30pm

SESSION 4: 2:30-3:45pm (3 slots)

• 2:30 - 2:55 Kamalika Chaudhuri: A Closer Look at Adversarial Robustness for Well-Separated Data
• 2:55 - 3:20 Changliu Liu: Run-time verification of deep neural networks
• 3:20 - 3:45 Clark Barrett: Towards Formal Verification of Deep Neural Networks and Mykel Kochenderfer: Safety validation through combining adaptive stress testing with neural network verification tools

Closing Remarks: 3:45-3:50pm