Robust Augmentation for Multivariate Time Series Classification

TL;DR

This paper demonstrates that simple augmentation techniques like cutout, cutmix, mixup, and window warp significantly enhance the robustness and accuracy of various neural network architectures in multivariate time series classification, especially with small datasets.

Contribution

It introduces and empirically validates the effectiveness of simple augmentation methods across multiple neural architectures for time series classification, improving performance on the UEA MTSC datasets.

Findings

Augmentation improves accuracy by 1% to 45% across 18 datasets.

All tested architectures benefit from augmentation techniques.

Augmentation methods are statistically significant in enhancing model robustness.

Abstract

Neural networks are capable of learning powerful representations of data, but they are susceptible to overfitting due to the number of parameters. This is particularly challenging in the domain of time series classification, where datasets may contain fewer than 100 training examples. In this paper, we show that the simple methods of cutout, cutmix, mixup, and window warp improve the robustness and overall performance in a statistically significant way for convolutional, recurrent, and self-attention based architectures for time series classification. We evaluate these methods on 26 datasets from the University of East Anglia Multivariate Time Series Classification (UEA MTSC) archive and analyze how these methods perform on different types of time series data.. We show that the InceptionTime network with augmentation improves accuracy by 1% to 45% in 18 different datasets compared to without augmentation. We also show that augmentation improves accuracy for recurrent and self attention based architectures.