Harmonic Dataset Distillation for Time Series Forecasting

TL;DR

Harmonic Dataset Distillation (HDT) is a novel method for time series forecasting that leverages frequency domain analysis to create compact datasets, improving scalability and generalization across architectures.

Contribution

HDT introduces a frequency domain approach to dataset distillation for time series, addressing overfitting and scalability issues in conventional methods.

Findings

HDT achieves strong cross-architecture generalization.

HDT scales effectively to large real-world datasets.

HDT maintains temporal dependencies during distillation.

Abstract

Time Series forecasting (TSF) in the modern era faces significant computational and storage cost challenges due to the massive scale of real-world data. Dataset Distillation (DD), a paradigm that synthesizes a small, compact dataset to achieve training performance comparable to that of the original dataset, has emerged as a promising solution. However, conventional DD methods are not tailored for time series and suffer from architectural overfitting and limited scalability. To address these issues, we propose Harmonic Dataset Distillation for Time Series Forecasting (HDT). HDT decomposes the time series into its sinusoidal basis through the FFT and aligns the core periodic structure by Harmonic Matching. Since this process operates in the frequency domain, all updates during distillation are applied globally without disrupting temporal dependencies of time series. Extensive experiments demonstrate that HDT achieves strong cross-architecture generalization and scalability, validating its practicality for large-scale, real-world applications.