Timezone: »

Active Learning with Safety Constraints
Romain Camilleri · Andrew Wagenmaker · Jamie Morgenstern · Lalit Jain · Kevin Jamieson

Thu Dec 01 02:00 PM -- 04:00 PM (PST) @ Hall J #712

Active learning methods have shown great promise in reducing the number of samples necessary for learning. As automated learning systems are adopted into real-time, real-world decision-making pipelines, it is increasingly important that such algorithms are designed with safety in mind. In this work we investigate the complexity of learning the best safe decision in interactive environments. We reduce this problem to a safe linear bandits problem, where our goal is to find the best arm satisfying certain (unknown) safety constraints. We propose an adaptive experimental design-based algorithm, which we show efficiently trades off between the difficulty of showing an arm is unsafe vs suboptimal. To our knowledge, our results are the first on best-arm identification in linear bandits with safety constraints. In practice, we demonstrate that this approach performs well on synthetic and real world datasets.

Author Information

Romain Camilleri (University of Washington)
Andrew Wagenmaker (University of Washington)
Jamie Morgenstern (U Washington)
Lalit Jain (University of Washington)
Kevin Jamieson (U Washington)

More from the Same Authors