Learning to Optimize with Confidence
Event details
| Date | 23.09.2016 |
| Hour | 10:15 › 11:15 |
| Speaker |
Prof. Andreas Krause, ETHZ Bio: Andreas Krause is an Associate Professor of Computer Science at ETH Zurich, where he leads the Learning & Adaptive Systems Group. Before that he was an Assistant Professor of Computer Science at Caltech. He received his Ph.D. in Computer Science from Carnegie Mellon University and his Diplom in Computer Science and Mathematics from TU Munich, Germany. He is a Microsoft Research Faculty Fellow and received an ERC Starting Investigator grant, an NSF CAREER award as well as several best paper awards at premier conferences and journals. |
| Location | |
| Category | Conferences - Seminars |
With the success of machine learning, we increasingly see learning algorithms make decisions in the real world. Often, however, this is in stark contrast to the classical train-test paradigm, since the learning algorithm affects the very data it must operate on. A central challenge is to trade exploration — collecting data for the sake of fitting better models — and exploitation — using the estimate to make decisions. In many applications, such as in robotics, exploration is a potentially dangerous proposition, as it requires experimenting with actions with unknown consequences. In this talk, I will formalize the problem of safe exploration as one of optimizing an unknown function subject to unknown constraints. Both objective and constraints are revealed through noisy experiments, and safety requires that no infeasible action is chosen at any point. I will present an approach that uses Bayesian inference over the objective and constraints, which -- under some regularity assumptions -- is guaranteed to converge to a natural notion of reachable optimum. I will also show experiments on safe automatic parameter tuning of a robotic platform.
Practical information
- Expert
- Free