Interpretable Learning for Travel Behaviours in Cyber-Physical-Social-Systems

Abstract

Interpretable learning is important for understanding human behavioral patterns in Cyber-Physical-Social-Systems (CPSS). It facilitates smart decision-makings of intelligent algorithms so that the management of such human-machine hybrid systems can be efficient and optimal. Unlike the big data driven transportation management, this paper introduces a new interpretable learning method using fuzzy logic to semantically extract travel behaviors. Computational experiments based on actual traffic data indicate that our method is able to generate explicit rules, and these rules can be used to predict traffic patterns very well.