Finding Short Signals in Long Irregular Time Series with Continuous-Time Attention Policy Networks
This addresses classification challenges in domains like healthcare where irregular measurements and long series with sparse signals are common, representing a novel method for a known bottleneck.
The paper tackles the problem of classifying long irregularly-sampled time series where only small portions are relevant, by proposing CAT, a model that explicitly seeks highly-relevant portions using reinforcement learning and localized modeling, and it outperforms ten state-of-the-art methods on synthetic and real data.
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.