Like Humans to Few-Shot Learning through Knowledge Permeation of Vision and Text

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Jia, Yuyu, Zhou, Qing, Huang, Wei, Gao, Junyu, Wang, Qi
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911884580487168
author Jia, Yuyu
Zhou, Qing
Huang, Wei
Gao, Junyu
Wang, Qi
author_facet Jia, Yuyu
Zhou, Qing
Huang, Wei
Gao, Junyu
Wang, Qi
contents Few-shot learning aims to generalize the recognizer from seen categories to an entirely novel scenario. With only a few support samples, several advanced methods initially introduce class names as prior knowledge for identifying novel classes. However, obstacles still impede achieving a comprehensive understanding of how to harness the mutual advantages of visual and textual knowledge. In this paper, we propose a coherent Bidirectional Knowledge Permeation strategy called BiKop, which is grounded in a human intuition: A class name description offers a general representation, whereas an image captures the specificity of individuals. BiKop primarily establishes a hierarchical joint general-specific representation through bidirectional knowledge permeation. On the other hand, considering the bias of joint representation towards the base set, we disentangle base-class-relevant semantics during training, thereby alleviating the suppression of potential novel-class-relevant information. Experiments on four challenging benchmarks demonstrate the remarkable superiority of BiKop. Our code will be publicly available.
format Preprint
id arxiv_https___arxiv_org_abs_2405_12543
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Like Humans to Few-Shot Learning through Knowledge Permeation of Vision and Text
Jia, Yuyu
Zhou, Qing
Huang, Wei
Gao, Junyu
Wang, Qi
Computer Vision and Pattern Recognition
Artificial Intelligence
Few-shot learning aims to generalize the recognizer from seen categories to an entirely novel scenario. With only a few support samples, several advanced methods initially introduce class names as prior knowledge for identifying novel classes. However, obstacles still impede achieving a comprehensive understanding of how to harness the mutual advantages of visual and textual knowledge. In this paper, we propose a coherent Bidirectional Knowledge Permeation strategy called BiKop, which is grounded in a human intuition: A class name description offers a general representation, whereas an image captures the specificity of individuals. BiKop primarily establishes a hierarchical joint general-specific representation through bidirectional knowledge permeation. On the other hand, considering the bias of joint representation towards the base set, we disentangle base-class-relevant semantics during training, thereby alleviating the suppression of potential novel-class-relevant information. Experiments on four challenging benchmarks demonstrate the remarkable superiority of BiKop. Our code will be publicly available.
title Like Humans to Few-Shot Learning through Knowledge Permeation of Vision and Text
topic Computer Vision and Pattern Recognition
Artificial Intelligence
url https://arxiv.org/abs/2405.12543