Sparse Transformer with Local and Seasonal Adaptation for Multivariate Time Series Forecasting

TL;DR

This paper introduces Dozer Attention, a sparse attention mechanism for multivariate time series forecasting that captures locality, seasonality, and global dependencies efficiently, leading to improved performance over state-of-the-art methods.

Contribution

The paper proposes the Dozer Attention mechanism with three sparse components and the Dozerformer framework, advancing multivariate time series forecasting by enhancing efficiency and accuracy.

Findings

Outperforms recent state-of-the-art methods on nine benchmark datasets.

Excluding some historical data does not reduce forecasting accuracy.

Significantly improves computational efficiency in time series forecasting.

Abstract

Transformers have achieved remarkable performance in multivariate time series(MTS) forecasting due to their capability to capture long-term dependencies. However, the canonical attention mechanism has two key limitations: (1) its quadratic time complexity limits the sequence length, and (2) it generates future values from the entire historical sequence. To address this, we propose a Dozer Attention mechanism consisting of three sparse components: (1) Local, each query exclusively attends to keys within a localized window of neighboring time steps. (2) Stride, enables each query to attend to keys at predefined intervals. (3) Vary, allows queries to selectively attend to keys from a subset of the historical sequence. Notably, the size of this subset dynamically expands as forecasting horizons extend. Those three components are designed to capture essential attributes of MTS data, including locality, seasonality, and global temporal dependencies. Additionally, we present the Dozerformer Framework, incorporating the Dozer Attention mechanism for the MTS forecasting task. We evaluated the proposed Dozerformer framework with recent state-of-the-art methods on nine benchmark datasets and confirmed its superior performance. The experimental results indicate that excluding a subset of historical time steps from the time series forecasting process does not compromise accuracy while significantly improving efficiency. Code is available at https://github.com/GRYGY1215/Dozerformer.