Learning to Control Safety-Critical Systems

Published on ● Video Link: https://www.youtube.com/watch?v=Jpw1JeZBu8M



Duration: 58:45
4,316 views
32


Adam Wierman (California Institute of Technology)
https://simons.berkeley.edu/node/22753
Structure of Constraints in Sequential Decision-Making

Making use of modern black-box AI tools such as deep reinforcement learning is potentially transformational for safety-critical systems such as data centers, the electricity grid, transportation, and beyond. However, such machine-learned algorithms typically do not have formal guarantees on their worst-case performance, stability, or safety and are typically difficult to make use of in distributed, networked settings. So, while their performance may improve upon traditional approaches in “typical” cases, they may perform arbitrarily worse in scenarios where the training examples are not representative due to, e.g., distribution shift, or in situations where global information is unavailable to local controllers. These represent significant drawbacks when considering the use of AI tools in safety-critical networked systems. Thus, a challenging open question emerges: Is it possible to provide guarantees that allow black-box AI tools to be used in safety-critical applications? In this talk, I will provide an overview of a variety of projects from my group that seek to develop robust and localizable tools combining model-free and model-based approaches to yield AI tools with formal guarantees on performance, stability, safety, and sample complexity.




Other Videos By Simons Institute for the Theory of Computing


2022-10-26Linear Growth of Quantum Circuit Complexity
2022-10-26Mathematics of the COVID-19 Pandemics: Lessons Learned and How to Mitigate the Next One
2022-10-25Efficient and Targeted COVID-19 Border Testing via Reinforcement Learning
2022-10-25Random Walks on Simplicial Complexes for Exploring Networks
2022-10-25Functional Law of Large Numbers and PDEs for Spatial Epidemic Models with...
2022-10-25Algorithms Using Local Graph Features to Predict Epidemics
2022-10-24Epidemic Models with Manual and Digital Contact Tracing
2022-10-21Pandora’s Box: Learning to Leverage Costly Information
2022-10-20Thresholds
2022-10-19NLTS Hamiltonians from Codes | Quantum Colloquium
2022-10-15Learning to Control Safety-Critical Systems
2022-10-14Near-Optimal No-Regret Learning for General Convex Games
2022-10-14The Power of Adaptivity in Representation Learning: From Meta-Learning to Federated Learning
2022-10-14When Matching Meets Batching: Optimal Multi-stage Algorithms and Applications
2022-10-13Optimal Learning for Structured Bandits
2022-10-13Dynamic Spatial Matching
2022-10-13New Results on Primal-Dual Algorithms for Online Allocation Problems With Applications to ...
2022-10-12Learning Across Bandits in High Dimension via Robust Statistics
2022-10-12Are Multicriteria MDPs Harder to Solve Than Single-Criteria MDPs?
2022-10-12A Game-Theoretic Approach to Offline Reinforcement Learning
2022-10-11The Statistical Complexity of Interactive Decision Making



Tags:
Simons Institute
theoretical computer science
UC Berkeley
Computer Science
Theory of Computation
Theory of Computing
Structure of Constraints in Sequential Decision-Making
Adam Wierman