Few-Shot Learning for Images Classification Considering Feature Variance

Lei Wang, Minhao Xue, Li Wang, Ming Cheng

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

Abstract

Few-shot learning aims to solve the problem of learning with limited samples. However, existing metric-based meta-learning models have the issue of insufficient attention to feature representation capability. Therefore, this paper introduces the feature variance loss during network training to learn more powerful features. Experimental results demonstrate that our method achieves good learning performance on the miniImageNet and tieredImageNet datasets.

Original languageEnglish
Title of host publication2023 3rd International Conference on Electronic Information Engineering and Computer, EIECT 2023
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages508-511
Number of pages4
ISBN (Electronic)9798350357707
DOIs
StatePublished - 2023
Event3rd International Conference on Electronic Information Engineering and Computer, EIECT 2023 - Hybrid, Shenzhen, China
Duration: 17 Nov 202319 Nov 2023

Publication series

Name2023 3rd International Conference on Electronic Information Engineering and Computer, EIECT 2023

Conference

Conference3rd International Conference on Electronic Information Engineering and Computer, EIECT 2023
Country/TerritoryChina
CityHybrid, Shenzhen
Period17/11/2319/11/23

Keywords

  • Few-shot learning
  • feature variance
  • image classification
  • meta-learning
  • metric learning

Fingerprint

Dive into the research topics of 'Few-Shot Learning for Images Classification Considering Feature Variance'. Together they form a unique fingerprint.

Cite this