Point process-based modeling and analysis of asynchronous event sequences

Abstract

Real-world interactions among multiple entities, such as user behaviors in social networks, job hunting and hopping, and diseases and their complications, often exhibit self-triggering and mutually-triggering patterns. For example, a tweet of a twitter user may trigger further responses from her friends. A disease of a patient may trigger other complications. Temporal point processes, especially Hawkes processes and correcting processes, have a capability to capture the triggering patterns quantitatively. This dissertation introduces basic concepts of point processes and proposes a series of cutting-edge modeling and learning techniques for practical applications. In particular, the Granger causality analysis of Hawkes processes, the clustering problem of event sequences, the combination of deep learning and point processes, the robust predictive learning of point processes from imperfect observations, and some interesting applications of data mining and computer vision are discussed.