Dissimilarity-Preserving Representation Learning for One-Class Time Series Classification

We propose to embed time series in a latent space where pairwise Euclidean distances (EDs) between samples are equal to pairwise dissimilarities in the original space, for a given dissimilarity measure. To this end, we use auto-encoder (AE) and encoder-only neural networks to learn elastic dissimilarity measures, e.g., dynamic time warping (DTW), that are central to time series classification (Bagnall et al., 2017). The learned representations are used in the context of one-class classification (Mauceri et al., 2020) on the datasets of UCR/UEA archive (Dau et al., 2019). Using a 1-nearest neighbor (1NN) classifier, we show that learned representations allow classification performance that is close to that of raw data, but in a space of substantially lower dimensionality. This implies substantial and compelling savings in terms of computational and storage requirements for nearest neighbor time series classification.

Automated summary

We propose a method to embed time series into a latent space where pairwise Euclidean distances equal pairwise dissimilarities in the original space. By using auto-encoder and encoder-only neural networks to learn elastic dissimilarity measures like dynamic time warping, we achieve classification performance close to that of raw data but with significantly lower dimensionality. This provides substantial savings in computational and storage requirements for nearest neighbor time series classification.