Controlled Information Fusion with Risk-Averse CVaR Social Sensors

TL;DR

This paper studies a multi-agent system where risk-averse social sensors perform Bayesian social learning, and a controller optimizes long-term objectives through discriminatory pricing, modeled as a POMDP, with structural results on optimal policies.

Contribution

It introduces a novel framework combining risk-averse social learning with control via discriminatory pricing, formulated as a POMDP, and derives structural properties of the optimal control policy.

Findings

Optimal pricing sequence is a super-martingale, decreasing on average over time.

Structural results for the optimal control policy as a function of risk-aversion.

Modeling of social sensors' Bayesian learning with CVaR risk measure.

Abstract

Consider a multi-agent network comprised of risk averse social sensors and a controller that jointly seek to estimate an unknown state of nature, given noisy measurements. The network of social sensors perform Bayesian social learning - each sensor fuses the information revealed by previous social sensors along with its private valuation using Bayes' rule - to optimize a local cost function. The controller sequentially modifies the cost function of the sensors by discriminatory pricing (control inputs) to realize long term global objectives. We formulate the stochastic control problem faced by the controller as a Partially Observed Markov Decision Process (POMDP) and derive structural results for the optimal control policy as a function of the risk-aversion factor in the Conditional Value-at-Risk (CVaR) cost function of the sensors. We show that the optimal price sequence when the sensors are risk- averse is a super-martingale; i.e, it decreases on average over time.