Active Learning with Safety Constraints
This work addresses safety constraints in active learning for real-time decision-making systems, representing an incremental advance by applying existing methods to a new safety-focused context.
The paper tackles the problem of learning the best safe decision in interactive environments by reducing it to a constrained linear bandits problem, and proposes an adaptive algorithm that efficiently trades off between safety and optimality, demonstrating good performance on synthetic and real-world datasets.
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 constrained 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.