DDTime: Dataset Distillation with Spectral Alignment and Information Bottleneck for Time-Series Forecasting

TL;DR

DDTime introduces a novel dataset distillation framework for time-series forecasting that uses spectral alignment and information bottleneck principles to improve synthetic data quality and model accuracy efficiently.

Contribution

The paper presents DDTime, a lightweight, plug-in distillation method that addresses autocorrelation bias and synthetic sample diversity in time-series data, enhancing distillation effectiveness.

Findings

Achieves about 30% relative accuracy improvement over existing methods.

Introduces spectral alignment to mitigate autocorrelation bias.

Adds only 2.49% computational overhead.

Abstract

Time-series forecasting is fundamental across many domains, yet training accurate models often requires large-scale datasets and substantial computational resources. Dataset distillation offers a promising alternative by synthesizing compact datasets that preserve the learning behavior of full data. However, extending dataset distillation to time-series forecasting is non-trivial due to two fundamental challenges: 1.temporal bias from strong autocorrelation, which leads to distorted value-term alignment between teacher and student models; and 2.insufficient diversity among synthetic samples, arising from the absence of explicit categorical priors to regularize trajectory variety. In this work, we propose DDTime, a lightweight and plug-in distillation framework built upon first-order condensation decomposition. To tackle Challenge 1, it revisits value-term alignment through temporal statistics and introduces a frequency-domain alignment mechanism to mitigate autocorrelation-induced bias, ensuring spectral consistency and temporal fidelity. To address Challenge 2, we further design an inter-sample regularization inspired by the information bottleneck principle, which enhances diversity and maximizes information density across synthetic trajectories. The combined objective is theoretically compatible with a wide range of condensation paradigms and supports stable first-order optimization. Extensive experiments on 20 benchmark datasets and diverse forecasting architectures demonstrate that DDTime consistently outperforms existing distillation methods, achieving about 30% relative accuracy gains while introducing about 2.49% computational overhead. All code and distilled datasets will be released.