Intensity-Free Learning of Temporal Point Processes

Abstract

Temporal point processes are the dominant paradigm for modeling sequences of events happening at irregular intervals. The standard way of learning in such models is by estimating the conditional intensity function. However, parameterizing the intensity function usually incurs several trade-offs. We show how to overcome the limitations of intensity-based approaches by directly modeling the conditional distribution of inter-event times. We draw on the literature on normalizing flows to design models that are flexible and efficient. We additionally propose a simple mixture model that matches the flexibility of flow-based models, but also permits sampling and computing moments in closed form. The proposed models achieve state-of-the-art performance in standard prediction tasks and are suitable for novel applications, such as learning sequence embeddings and imputing missing data.

arXiv commentInternational Conference on Learning Representations (ICLR) 2020

References in corpus (12)

• WaveNet: A Generative Model for Raw Audio
• Categorical Reparameterization with Gumbel-Softmax
• Variational Inference with Normalizing Flows
• REBAR: Low-variance, unbiased gradient estimates for discrete latent variable models
• Monte Carlo Gradient Estimation in Machine Learning
• Fully Neural Network based Model for General Temporal Point Processes
• Deep Reinforcement Learning of Marked Temporal Point Processes
• Learning Temporal Point Processes via Reinforcement Learning
• Latent Normalizing Flows for Discrete Sequences
• Sum-of-Squares Polynomial Flow
• Distilling Information Reliability and Source Trustworthiness from Digital Traces
• Uncertainty on Asynchronous Time Event Prediction

Cited by in corpus (5)

• Spatio-Temporal Graph Neural Point Process for Traffic Congestion Event Prediction
• Self-Attentive Hawkes Processes
• Neural Hybrid Automata: Learning Dynamics with Multiple Modes and Stochastic Transitions
• Neural Temporal Point Processes For Modelling Electronic Health Records
• Neural Spatio-Temporal Point Processes