Adaptive Sensor Placement for Continuous Spaces

Date August 25, 2020
Authors James Grant & David Leslie (Lancaster University), Alexis Boukouvalas, Ryan-Rhys Griffiths (University of Cambridge), Sattar Vakili, Enrique Munoz De Cote

Repeatedly placing sensors along an interval (in time or space) to detect as many events as possible. Each replacement opportunity needs to balance exploration (finding out information for future deployments) and exploitation (detecting events in the current deployment). The difference from standard bandit papers is that the actions are intervals. There are obvious generalisations to selecting regions instead of intervals.

View the paper

Share
,,

Related articles

Soft Q-Learning with Mutual-Information Regularization

Secondmind appoints Professor Carl Edward Rasmussen as Chief Scientist