Self-reconstruction network for fine-grained few-shot classification

Li, X., Li, Z., Xie, J., Yang, X. , Xue, J.-H. and Ma, Z. (2024) Self-reconstruction network for fine-grained few-shot classification. Pattern Recognition, (doi: 10.1016/j.patcog.2024.110485) (Early Online Publication)

Abstract

Metric-based methods are one of the most common methods to solve the problem of few-shot image classification. However, traditional metric-based few-shot methods suffer from overfitting and local feature misalignment. The recently proposed feature reconstruction-based approach, which reconstructs query image features from the support set features of a given class and compares the distance between the original query features and the reconstructed query features as the classification criterion, effectively solves the feature misalignment problem. However, the issue of overfitting still has not been considered. To this end, we propose a self-reconstruction metric module for diversifying query features and a restrained cross-entropy loss for avoiding over-confident predictions. By introducing them, the proposed self-reconstruction network can effectively alleviate overfitting. Extensive experiments on five benchmark fine-grained datasets demonstrate that our proposed method achieves state-of-the-art performance on both 5-way 1-shot and 5-way 5-shot classification tasks.

• Item Type: Articles
• Additional Information: This work was supported in part by the Beijing Natural Science Foundation Project No. Z200002, in part by the Royal Society under International Exchanges Award IEC\NSFC\201071, in part by the National Natural Science Foundation of China (NSFC) No. 62111530146, 62176110, 61906080, 61922015, U19B2036, 62225601, in part by Young Doctoral Fund of Education Department of Gansu Province under Grant 2021QB-038, Youth Innovative Research Team of BUPT No. 2023QNTD02, and Hong-liu Distinguished Young Talents Foundation of Lanzhou University of Technology .
• Keywords: Few-shot learning, fine-grained image classification, deep neural network, self-reconstruction network.
• Status: Early Online Publication
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: Yang, Dr Xiaochen
• Authors: Li, X., Li, Z., Xie, J., Yang, X., Xue, J.-H., and Ma, Z.
• College/School: College of Science and Engineering > School of Mathematics and Statistics > Statistics
• Journal Name: Pattern Recognition
• Publisher: Elsevier
• ISSN: 0031-3203
• ISSN (Online): 1873-5142
• Published Online: 06 April 2024
• Copyright Holders: Copyright © 2024 The Authors
• First Published: First published in Pattern Recognition 2024
• Publisher Policy: Reproduced under a Creative Commons licence