Home > Research > Publications & Outputs > Adaptive sensor placement for continuous spaces

Electronic data

  • 1905.06821

    Accepted author manuscript, 2 MB, PDF document

    Available under license: CC BY-NC: Creative Commons Attribution-NonCommercial 4.0 International License

Links

View graph of relations

Adaptive sensor placement for continuous spaces

Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSNConference contribution/Paper

Published
Close
Publication date12/07/2019
Host publicationProceedings of the 36 th International Conference on Machine Learning, Long Beach, California, PMLR 97, 2019
PublisherProceedings of Machine Learning Research
Number of pages9
Original languageEnglish

Publication series

NameProceedings of Machine Learning Research
Volume97
ISSN (Print)2640-3498

Abstract

We consider the problem of adaptively placing sensors along an interval to detect stochasticallygenerated events. We present a new formulation of the problem as a continuum-armed bandit problem with feedback in the form of partial observations of realisations of an inhomogeneous Poisson process. We design a solution method by combining Thompson sampling with nonparametric inference via increasingly granular Bayesian histograms and derive an O˜(T2/3) bound on the Bayesian regret in T rounds. This is coupled with the design of an efficent optimisation approach to
select actions in polynomial time. In simulations we demonstrate our approach to have substantially lower and less variable regret than competitor algorithms.