DA-Net: Dual-attention network for multivariate time series classification

Multivariate time series classification is one of the increasingly important issues in machine learning. Existing methods focus on establishing the global long-range dependen-cies or discovering the local critical sequence fragments. However, they often ignore the combined information from both global and local features. In this paper, we propose a novel network (called DA-Net) based on dual attention to mine the local???global features for multivariate time series classification. Specifically, DA-Net consists of two distinctive layers, i.e., the Squeeze-Excitation Window Attention (SEWA) layer and the Sparse Self -Attention within Windows (SSAW) layer. For the SEWA layer, we capture the local window-wise information by explicitly establishing window dependencies to prioritize critical windows. For the SSAW layer, we preserve rich activate scores with less computa-tion to widen the window scope for capturing global long-range dependencies. Based on the two elaborated layers, DA-Net can mine critical local sequence fragments in the process of establishing global long-range dependencies. The experimental results show that DA -Net is able to achieve competing performance with state-of-the-art approaches on the mul-tivariate time series classification. ?? 2022 Elsevier Inc. All rights reserved.

Automated summary

In this paper, a novel network called DA-Net is proposed for mining local-global features in multivariate time series classification. DA-Net consists of two layers, the Squeeze-Excitation Window Attention (SEWA) layer and the Sparse Self-Attention within Windows (SSAW) layer, which capture local and global features and mine critical local sequence fragments through establishing global long-range dependencies.