Details

Autor(en) / Beteiligte

• Gao, Yi
• Feng, Wentao
• Xiao, Rong
• He, Lihuo
• He, Zhenan
• Lv, Jiancheng
• Tang, Chenwei

Titel

Improving generalized zero-shot learning via cluster-based semantic disentangling representation

Ist Teil von

• Pattern recognition, 2024-06, Vol.150, p.110320, Article 110320

Ort / Verlag

Elsevier Ltd

Erscheinungsjahr

2024

Quelle

Alma/SFX Local Collection

Beschreibungen/Notizen

• Generalized Zero-Shot Learning (GZSL) aims to recognize both seen and unseen classes by training only the seen classes, in which the instances of unseen classes tend to be biased towards the seen class. In this paper, we propose a Cluster-based Semantic Disentangling Representation (CSDR) method to improve GZSL by alleviating the problems of domain shift and semantic gap. First, we cluster the seen data into multiple clusters, where the samples in each cluster belong to several original seen categories, so as to facilitate fine-grained semantic disentangling of visual feature vectors. Then, we introduce representation random swapping and contrastive learning based on the clustering results to realize the disentangling semantic representations of semantic-unspecific, class-shared, and class-unique. The fine-grained semantic disentangling representations show high intra-class similarity and inter-class discriminability, which improve the performance of GZSL by alleviating the problem of domain shift. Finally, we construct the visual-semantic embedding space by the variational auto-encoder and alignment module, which can bridge the semantic gap by generating strongly discriminative unseen class samples. Extensive experimental results on four public data sets prove that our method significantly outperforms state-of-the-art methods in generalized and conventional settings. •The proposed CSDR method improves GZSL by alleviating domain shift and semantic gap.•Semantic disentangling module learns fine-grained semantic feature of visual vectors.•Semantic representation module rises intra-class similarity & inter-discernibility.•Visual-semantic embedding module transfers knowledge from seen class to unseen class.•Experiments on four benchmarks prove effectiveness and superiority of CSDR method.