Enhancing Asynchronous Time Series Forecasting with Contrastive Relational Inference

TL;DR

This paper introduces CRIHP, a novel model that uses neural relational inference and contrastive learning to explicitly capture event interactions in asynchronous time series, improving prediction accuracy.

Contribution

It proposes a new approach combining neural relational inference with contrastive learning within a variational framework for better event interaction modeling.

Findings

CRIHP outperforms existing models on real-world datasets

The model effectively captures complex event interactions

Code will be integrated into EasyTPP framework

Abstract

Asynchronous time series, also known as temporal event sequences, are the basis of many applications throughout different industries. Temporal point processes(TPPs) are the standard method for modeling such data. Existing TPP models have focused on parameterizing the conditional distribution of future events instead of explicitly modeling event interactions, imposing challenges for event predictions. In this paper, we propose a novel approach that leverages Neural Relational Inference (NRI) to learn a relation graph that infers interactions while simultaneously learning the dynamics patterns from observational data. Our approach, the Contrastive Relational Inference-based Hawkes Process (CRIHP), reasons about event interactions under a variational inference framework. It utilizes intensity-based learning to search for prototype paths to contrast relationship constraints. Extensive experiments on three real-world datasets demonstrate the effectiveness of our model in capturing event interactions for event sequence modeling tasks. Code will be integrated into the EasyTPP framework.