Files
Abstract
Temporal data are continuously collected in a wide range of domains. Time series classification, as an essential task in time series analysis, aims to assign a set of temporal sequences to different categories. Time series classification using distance metric learning such as the virtual sequence metric learning (VSML), have achieved remarkable performance, where virtual sequences attract samples from different classes to facilitate classification. However, the existing VSML methods simply employ fixed virtual sequences, which might not be optimal for the subsequent classification tasks. To address this issue, we propose a novel time series classification method, discriminative virtual sequence learning (DVSL). Following the framework of sequence metric learning, our DVSL method jointly learns a set of discriminative virtual sequences that help separate time series samples in a feature space and optimizes the temporal alignment by DTW measure. Extensive experiments on UCR datasets demonstrate the efficiency of DVSL, compared to several baselines.