Active Exploration via Experiment Design in Markov Chains
This work addresses the challenge of active exploration for applications like reinforcement learning and spatial monitoring, presenting a novel algorithmic approach with theoretical guarantees, but it appears incremental as it builds on classical experimental design concepts.
The paper tackles the problem of designing experiments in a Markov chain setting where experiments are associated with states and policies control transitions, proposing an algorithm called Markov-design that provably converges to optimal measurement allocation. The result is demonstrated with applications in ecological surveillance and pharmacology, though no concrete performance numbers are provided.
A key challenge in science and engineering is to design experiments to learn about some unknown quantity of interest. Classical experimental design optimally allocates the experimental budget to maximize a notion of utility (e.g., reduction in uncertainty about the unknown quantity). We consider a rich setting, where the experiments are associated with states in a {\em Markov chain}, and we can only choose them by selecting a {\em policy} controlling the state transitions. This problem captures important applications, from exploration in reinforcement learning to spatial monitoring tasks. We propose an algorithm -- \textsc{markov-design} -- that efficiently selects policies whose measurement allocation \emph{provably converges to the optimal one}. The algorithm is sequential in nature, adapting its choice of policies (experiments) informed by past measurements. In addition to our theoretical analysis, we showcase our framework on applications in ecological surveillance and pharmacology.