Generative Wavelet-Multilayer Perception Feature Fusion Method for Zero-Shot Learning

Qun Xiong Zhu, Zi Shu Gao, Ning Zhang, Yan Lin He, Yuan Xu

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Citation (Scopus)

Abstract

For zero-shot learning, a Generative WaveletMultilayer Perception Feature Fusion (GWMPFF) method is proposed. In GWMPFF, the Generative Adversarial Networks(GAN) is used to generate virtual samples from unseen classes to alleviate the domain shift problem. Considering that the sensitivity of wavelet transform to contour information and high-frequency details is consistent with the characteristics of semantic description, a WaveletMultilayer Perceptron Feature Fusion (WMPFF) model is proposed to transform visual samples into semantic descriptions for classification. Finally, experiments on the standard datasets SUN and CUB demonstrate that the proposed method outperforms the comparison methods.

Original languageEnglish
Title of host publicationProceeding - 2021 China Automation Congress, CAC 2021
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1233-1238
Number of pages6
ISBN (Electronic)9781665426473
DOIs
Publication statusPublished - 2021
Externally publishedYes
Event2021 China Automation Congress, CAC 2021 - Beijing, China
Duration: 22 Oct 202124 Oct 2021

Publication series

NameProceeding - 2021 China Automation Congress, CAC 2021

Conference

Conference2021 China Automation Congress, CAC 2021
Country/TerritoryChina
CityBeijing
Period22/10/2124/10/21

Keywords

  • Feature fusion
  • Generative adversarial networks
  • Wavelet transform
  • Zero-shot learning

Fingerprint

Dive into the research topics of 'Generative Wavelet-Multilayer Perception Feature Fusion Method for Zero-Shot Learning'. Together they form a unique fingerprint.

Cite this