Breaking the Context Bottleneck on Long Time Series Forecasting

TL;DR

This paper introduces the Logsparse Decomposable Multiscaling (LDM) framework, which improves long-term time series forecasting by effectively leveraging multiscale patterns, reducing overfitting, and enhancing efficiency, outperforming existing methods.

Contribution

The paper proposes the LDM framework that decouples multiscale patterns in time series, leading to better long-term forecasting, efficiency, and architectural simplicity compared to prior approaches.

Findings

LDM outperforms all baseline models on long-term forecasting benchmarks.

LDM reduces training time and memory costs.

Decoupling multiscale patterns improves forecast accuracy and efficiency.

Abstract

Long-term time-series forecasting is essential for planning and decision-making in economics, energy, and transportation, where long foresight is required. To obtain such long foresight, models must be both efficient and effective in processing long sequence. Recent advancements have enhanced the efficiency of these models; however, the challenge of effectively leveraging longer sequences persists. This is primarily due to the tendency of these models to overfit when presented with extended inputs, necessitating the use of shorter input lengths to maintain tolerable error margins. In this work, we investigate the multiscale modeling method and propose the Logsparse Decomposable Multiscaling (LDM) framework for the efficient and effective processing of long sequences. We demonstrate that by decoupling patterns at different scales in time series, we can enhance predictability by reducing non-stationarity, improve efficiency through a compact long input representation, and simplify the architecture by providing clear task assignments. Experimental results demonstrate that LDM not only outperforms all baselines in long-term forecasting benchmarks, but also reducing both training time and memory costs.