Center-VAE with discriminative and semantic-relevant fine-tuning features for generalized zero-shot learning

Zhai, Zhibo; Li, Xiao; Chang, Zhonghao

doi:10.1016/j.image.2022.116897

Cited by 5 publications

(2 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Generating synthesized visual features alleviates the classification bias to a certain extent, but there still exist gaps among synthesized features and high-quality features, such as Redundancy-Free Feature-based Generalized Zero-Shot Learning (RFF-GZSL) [11], TF-VAEGAN [26], SE-GZSL [18], and FREE [3]. Besides, in comparison with the recent TDCSS [6], CMPN [10], DFTN [17], CvDSF [40], CMC-GAN [38], and SALN [39], Co-GZSL still achieves excellent improvements on the performance. Moreover, although the results of CE-GZSL [12] are close to our model, it exists a gap between them.…”

Section: Comparison With State-of-the-arts (Sotas)mentioning

confidence: 99%

“…However, these methods tend to recognize unseen classes as seen classes in the context of lacking training samples on unseen classes in GZSL tasks, which results in low accuracy during testing. Therefore, recent works propose generativebased models, such as the Generative Dual Adversarial Network (GDAN) [16], the GZSL via Synthesized Examples (SE-GZSL) [32], Inference guided Feature Generation (Inf-FG) [13], Cross-Modal ConsistencyGAN (CMC-GAN) [38], a Generative Adversarial Network (GAN) named f-CLSWGAN [36], center-VAE with discriminative and semantic-relevant fine-tuning features for generalized zero-shot learning (CvDSF) [40], and Dual-Focus Transfer Network (DFTN) [17], to synthesize the sufficient number of samples for the unseen classes. Semantic features are usually mapped as synthesized visual features in generative models to tackle the problem that the unseen classes lack enough training samples.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Co-GZSL: Feature Contrastive Optimization for Generalized Zero-Shot Learning

Li,

Zhan,

Shen

et al. 2024

Neural Process Lett

View full text Add to dashboard Cite

Generalized Zero-Shot Learning (GZSL) learns from only labeled seen classes during training but discriminates both seen and unseen classes during testing. In GZSL tasks, most of the existing methods commonly utilize visual and semantic features for training. Due to the lack of visual features for unseen classes, recent works generate real-like visual features by using semantic features. However, the synthesized features in the original feature space lack discriminative information. It is important that the synthesized visual features should be similar to the ones in the same class, but different from the other classes. One way to solve this problem is to introduce the embedding space after generating visual features. Following this situation, the embedded features from the embedding space can be inconsistent with the original semantic features. For another way, some recent methods constrain the representation by reconstructing the semantic features using the original visual features and the synthesized visual features. In this paper, we propose a hybrid GZSL model, named feature Contrastive optimization for GZSL (Co-GZSL), to reconstruct the semantic features from the embedded features, which ensures that the embedded features are close to the original semantic features indirectly by comparing reconstructed semantic features with original semantic features. In addition, to settle the problem that the synthesized features lack discrimination and semantic consistency, we introduce a Feature Contrastive Optimization Module (FCOM) and jointly utilize contrastive and semantic cycle-consistency losses in the FCOM to strengthen the intra-class compactness and the inter-class separability and to encourage the model to generate semantically consistent and discriminative visual features. By combining the generative module, the embedding module, and the FCOM, we achieve Co-GZSL. We evaluate the proposed Co-GZSL model on four benchmarks, and the experimental results indicate that our model is superior over current methods. Code is available at: https://github.com/zhanzhuxi/Co-GZSL.

show abstract