Don't overfit the history -- Recursive time series data augmentation

TL;DR

This paper introduces RIM, a recursive data augmentation method for time series that generates new samples preserving original dynamics, reducing overfitting and improving model generalization across various tasks.

Contribution

The paper proposes a novel recursive interpolation framework for time series augmentation that maintains inherent dynamics and enhances learning performance.

Findings

RIM improves generalization in regression, classification, and reinforcement learning tasks.

Theoretical guarantees support the effectiveness of RIM.

Empirical results show significant performance gains over non-augmented data.

Abstract

Time series observations can be seen as realizations of an underlying dynamical system governed by rules that we typically do not know. For time series learning tasks, we need to understand that we fit our model on available data, which is a unique realized history. Training on a single realization often induces severe overfitting lacking generalization. To address this issue, we introduce a general recursive framework for time series augmentation, which we call Recursive Interpolation Method, denoted as RIM. New samples are generated using a recursive interpolation function of all previous values in such a way that the enhanced samples preserve the original inherent time series dynamics. We perform theoretical analysis to characterize the proposed RIM and to guarantee its test performance. We apply RIM to diverse real world time series cases to achieve strong performance over non-augmented data on regression, classification, and reinforcement learning tasks.