Decentralized decision making with spatially distributed data

I will present a number of problems that arise in learning and decision-making with spatially distributed data, such as those collected and processed by a sensor network. I will discuss two key issues in such problems. The first issue involves the aggregation of data from individual
sensors while taking into account the statistical spatial dependency of sensor measurements, and the communication constraints of the network. The second issue involves accounting for the spatial dependency in making simultaneous decisions for each of the sensors. I will highlight these issues in several problems of decentralized detection, and discuss some initial progress.