Modeling Temporal Dependencies within the Target for Long-Term Time Series Forecasting

TL;DR

This paper introduces TDAlign, a flexible and efficient framework that improves long-term time series forecasting by better modeling temporal dependencies within the target, leading to significant accuracy gains across multiple datasets.

Contribution

The paper proposes a novel plug-and-play framework, TDAlign, that enhances existing LTSF methods with minimal overhead by aligning change values and adaptively balancing loss functions.

Findings

Reduces baseline prediction errors by up to 9.19%.

Decreases change value errors by up to 15.78%.

Demonstrates effectiveness across six baselines and seven datasets.

Abstract

Long-term time series forecasting (LTSF) is a critical task across diverse domains. Despite significant advancements in LTSF research, we identify a performance bottleneck in existing LTSF methods caused by the inadequate modeling of Temporal Dependencies within the Target (TDT). To address this issue, we propose a novel and generic temporal modeling framework, Temporal Dependency Alignment (TDAlign), that equips existing LTSF methods with TDT learning capabilities. TDAlign introduces two key innovations: 1) a loss function that aligns the change values between adjacent time steps in the predictions with those in the target, ensuring consistency with variation patterns, and 2) an adaptive loss balancing strategy that seamlessly integrates the new loss function with existing LTSF methods without introducing additional learnable parameters. As a plug-and-play framework, TDAlign enhances existing methods with minimal computational overhead, featuring only linear time complexity and constant space complexity relative to the prediction length. Extensive experiments on six strong LTSF baselines across seven real-world datasets demonstrate the effectiveness and flexibility of TDAlign. On average, TDAlign reduces baseline prediction errors by \textbf{1.47\%} to \textbf{9.19\%} and change value errors by \textbf{4.57\%} to \textbf{15.78\%}, highlighting its substantial performance improvements.