The Neural Hawkes Process: A Neurally Self-Modulating Multivariate Point Process

Abstract

Many events occur in the world. Some event types are stochastically excited or inhibited---in the sense of having their probabilities elevated or decreased---by patterns in the sequence of previous events. Discovering such patterns can help us predict which type of event will happen next and when. We model streams of discrete events in continuous time, by constructing a neurally self-modulating multivariate point process in which the intensities of multiple event types evolve according to a novel continuous-time LSTM. This generative model allows past events to influence the future in complex and realistic ways, by conditioning future event intensities on the hidden state of a recurrent neural network that has consumed the stream of past events. Our model has desirable qualitative properties. It achieves competitive likelihood and predictive accuracy on real and synthetic datasets, including under missing-data conditions.

NIPS 2017 camera-ready. New experiments including intensity prediction evaluation, sensitivity to # of parameters, training speed analysis. Results updated to use final test data instead of devtest. Improved exposition, especially of continuous-time LSTM and thinning algorithm

References in corpus (4)

• Visualizing and Understanding Recurrent Networks
• Learning Granger Causality for Hawkes Processes
• Learning Network of Multivariate Hawkes Processes: A Time Series Approach
• Joint Modeling of Event Sequence and Time Series with Attentional Twin Recurrent Neural Networks

Cited by in corpus (23)

• Deep Learning for Time Series Forecasting: Tutorial and Literature Survey
• Foundations and modelling of dynamic networks using Dynamic Graph Neural Networks: A survey
• TREND: TempoRal Event and Node Dynamics for Graph Representation Learning
• Spatio-Temporal Graph Neural Point Process for Traffic Congestion Event Prediction
• Visual Causality Analysis of Event Sequence Data
• Learning Hawkes Processes from Short Doubly-Censored Event Sequences
• Learning to Simulate Daily Activities via Modeling Dynamic Human Needs
• TIGGER: Scalable Generative Modelling for Temporal Interaction Graphs
• Early Rumor Detection Using Neural Hawkes Process with a New Benchmark Dataset
• Ask "Who", Not "What": Bitcoin Volatility Forecasting with Twitter Data
• Modeling of Spatio-Temporal Hawkes Processes with Randomized Kernels
• Identifying Cost-effective Debunkers for Multi-stage Fake News Mitigation Campaigns
• Long Horizon Forecasting With Temporal Point Processes
• Recurrent neural network based parameter estimation of Hawkes model on high-frequency financial data
• Modeling Event Propagation via Graph Biased Temporal Point Process
• Generalized Evolutionary Point Processes: Model Specifications and Model Comparison
• A Graph Regularized Point Process Model For Event Propagation Sequence
• Modelling Direct Messaging Networks with Multiple Recipients for Cyber Deception
• Popularity Prediction for Social Media over Arbitrary Time Horizons
• Recurrent Point Review Models
• Event-Based Limit Order Book Simulation under a Neural Hawkes Process: Application in Market-Making
• Semi-supervised Learning for Marked Temporal Point Processes
• Provable local learning rule by expert aggregation for a Hawkes network