Adversarial Gaussian Process Regression In Sensor Network

BIO

Yi Li is a PhD student in Computer Science at the Vanderbilt University, EECS Department. Please email yi.li@vanderbilt.edu to contact Yi.

ABSTRACT

In this work, we consider machine learning models for anomaly detection based on Gaussian process regression and make the following contributions: (1) We define stealthy attacks based on optimization approaches and investigate the feasibility of designing undetectable attacks with catastrophic potential damage; (2) We model the interactions between the attacker and the defender as a Stackelberg game and design algorithms for computing the equilibrium, and (3) design resilient anomaly detectors for stealthy attacks based on the game theoretical framework.