Please use this identifier to cite or link to this item:
Title: Hierarchical domain adaptation with local feature patterns
Authors: Wen, Jun
Yuan, Junsong
Zheng, Qian
Liu, Risheng
Gong, Zhefeng
Zheng, Nenggan
Keywords: Engineering::Electrical and electronic engineering
Issue Date: 2022
Source: Wen, J., Yuan, J., Zheng, Q., Liu, R., Gong, Z. & Zheng, N. (2022). Hierarchical domain adaptation with local feature patterns. Pattern Recognition, 124, 108445-.
Journal: Pattern Recognition
Abstract: Domain adaptation is proposed to generalize learning machines and address performance degradation of models that are trained from one specific source domain but applied to novel target domains. Existing domain adaptation methods focus on transferring holistic features whose discriminability is generally tailored to be source-specific and inferiorly generic to be transferable. As a result, standard domain adaptation on holistic features usually damages feature structures, especially local feature statistics, and deteriorates the learned discriminability. To alleviate this issue, we propose to transfer primitive local feature patterns, whose discriminability are shown to be inherently more sharable, and perform hierarchical feature adaptation. Concretely, we first learn a cluster of domain-shared local feature patterns and partition the feature space into cells. Local features are adaptively aggregated inside each cell to obtain cell features, which are further integrated into holistic features. To achieve fine-grained adaptations, we simultaneously perform alignment on local features, cell features and holistic features, within which process the local and cell features are aligned independently inside each cell to maintain the learned local structures and prevent negative transfer. Experimenting on typical one-to-one unsupervised domain adaptation for both image classification and action recognition tasks, partial domain adaptation, and domain-agnostic adaptation, we show that the proposed method achieves more reliable feature transfer by consistently outperforming state-of-the-art models and the learned domain-invariant features generalize well to novel domains.
ISSN: 0031-3203
DOI: 10.1016/j.patcog.2021.108445
Schools: School of Electrical and Electronic Engineering 
Rights: © 2021 Elsevier Ltd. All rights reserved.
Fulltext Permission: none
Fulltext Availability: No Fulltext
Appears in Collections:EEE Journal Articles

Citations 20

Updated on Mar 1, 2024

Web of ScienceTM
Citations 20

Updated on Oct 31, 2023

Page view(s)

Updated on Mar 4, 2024

Google ScholarTM




Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.