Learning Hawkes Processes from Short Doubly-Censored Event Sequences

Abstract

Many real-world applications require robust algorithms to learn point processes based on a type of incomplete data --- the so-called short doubly-censored (SDC) event sequences. We study this critical problem of quantitative asynchronous event sequence analysis under the framework of Hawkes processes by leveraging the idea of data synthesis. Given SDC event sequences observed in a variety of time intervals, we propose a sampling-stitching data synthesis method --- sampling predecessors and successors for each SDC event sequence from potential candidates and stitching them together to synthesize long training sequences. The rationality and the feasibility of our method are discussed in terms of arguments based on likelihood. Experiments on both synthetic and real-world data demonstrate that the proposed data synthesis method improves learning results indeed for both time-invariant and time-varying Hawkes processes.

References in corpus (2)

• The Neural Hawkes Process: A Neurally Self-Modulating Multivariate Point Process
• Learning Granger Causality for Hawkes Processes

Cited by in corpus (9)

• Neural Jump Stochastic Differential Equations
• Interval-censored Hawkes processes
• PoPPy: A Point Process Toolbox Based on PyTorch
• Learning Registered Point Processes from Idiosyncratic Observations
• Event Outlier Detection in Continuous Time
• Estimating Network Structure from Incomplete Event Data
• Superposition-Assisted Stochastic Optimization for Hawkes Processes
• ABC Learning of Hawkes Processes with Missing or Noisy Event Times
• Adversarial Self-Paced Learning for Mixture Models of Hawkes Processes