Finding Short Signals in Long Irregular Time Series with Continuous-Time Attention Policy Networks

TL;DR

This paper introduces CAT, a novel model that effectively identifies and classifies short, relevant signals within long, irregular time series by explicitly focusing on key moments, outperforming existing methods.

Contribution

The paper presents a new reinforcement learning-based approach, CAT, that explicitly seeks relevant segments in long irregular time series for improved classification accuracy.

Findings

CAT outperforms ten state-of-the-art methods on synthetic and real datasets.

The model effectively identifies short signals within long, irregular time series.

Experimental results demonstrate superior classification performance with CAT.

Abstract

Irregularly-sampled time series (ITS) are native to high-impact domains like healthcare, where measurements are collected over time at uneven intervals. However, for many classification problems, only small portions of long time series are often relevant to the class label. In this case, existing ITS models often fail to classify long series since they rely on careful imputation, which easily over- or under-samples the relevant regions. Using this insight, we then propose CAT, a model that classifies multivariate ITS by explicitly seeking highly-relevant portions of an input series' timeline. CAT achieves this by integrating three components: (1) A Moment Network learns to seek relevant moments in an ITS's continuous timeline using reinforcement learning. (2) A Receptor Network models the temporal dynamics of both observations and their timing localized around predicted moments. (3) A recurrent Transition Model models the sequence of transitions between these moments, cultivating a representation with which the series is classified. Using synthetic and real data, we find that CAT outperforms ten state-of-the-art methods by finding short signals in long irregular time series.