Classifier ensemble based source-free domain adaptation for time series classification

doi:10.1016/j.knosys.2025.114584

Classifier ensemble based source-free domain adaptation for time series classification

2025-10-05

0

PRE

AI

DOI：10.1016/j.knosys.2025.114584

Original

Original request for help

Save

Abstract

En 中文

• Propose a novel CE-SFDA framework to addresses domain shift in time series classification task. • Design an ensemble classifier-based pseudo-labels generation method to improve the reliability of the pseudo-labels. • A memory-aware knowledge distillation method is introduced to mine both the global and local structural information of the target domain sample space. • An information entropy-based self-supervised learning strategy is introduced to align source and target domain distributions.

Classifier ensemble based source-free domain adaptation for time series classification

Abstract

Journal

Organization