Change point analysis for intelligent agents in city traffic

Abstract

Change point (CP) detection is an important problem in data mining (DM) applications. We consider this problem solving in multi-agent systems (MAS) domains. Change point testing allows agents to recognize changes in the environment, to detect more accurately current state information and provide more appropriate information for decision-making. Standard statistical procedures for change point detection, based on maximum likelihood estimators, are complex and require construction of parametrical models of data. In methods of computational statistics, such as bootstrapping or resampling, complex statistical inference is replaced by a large computation volumes. However, these methods require accurate analysis of their precision. In this paper, we apply and analyze a bootstrap-based CUSUM test for change point detection, as well as propose a pairwise resampling CP test. We derive some useful properties of the tests and demonstrate their application in the decentralized decision-making of vehicle agents in city traffic.