Bi-Branching Feature Interaction Representation Learning for Multivariate Time Series

Published: 01 Jan 2024, Last Modified: 11 Apr 2025Appl. Soft Comput. 2024EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Point out that existing methods fail to learn the timing characterization completely.•Inter and intra-variable features are equally crucial for representing time series.•Bi-FI model is designed to simultaneously learn inter and intra-variable features.•Bi-FI achieves SOTA performance in three mainstream time series analysis tasks.
Loading