Saved in:
Bibliographic Details
Main Authors: Lin, Guoliang, Xu, Yongheng, Lai, Hanjiang, Yin, Jian
Format: Preprint
Published: 2022
Subjects:
Online Access:https://arxiv.org/abs/2209.13816
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914786117156864
author Lin, Guoliang
Xu, Yongheng
Lai, Hanjiang
Yin, Jian
author_facet Lin, Guoliang
Xu, Yongheng
Lai, Hanjiang
Yin, Jian
contents Few-shot learning with $N$-way $K$-shot scheme is an open challenge in machine learning. Many metric-based approaches have been proposed to tackle this problem, e.g., the Matching Networks and CLIP-Adapter. Despite that these approaches have shown significant progress, the mechanism of why these methods succeed has not been well explored. In this paper, we try to interpret these metric-based few-shot learning methods via causal mechanism. We show that the existing approaches can be viewed as specific forms of front-door adjustment, which can alleviate the effect of spurious correlations and thus learn the causality. This causal interpretation could provide us a new perspective to better understand these existing metric-based methods. Further, based on this causal interpretation, we simply introduce two causal methods for metric-based few-shot learning, which considers not only the relationship between examples but also the diversity of representations. Experimental results demonstrate the superiority of our proposed methods in few-shot classification on various benchmark datasets. Code is available in https://github.com/lingl1024/causalFewShot.
format Preprint
id arxiv_https___arxiv_org_abs_2209_13816
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle Revisiting Few-Shot Learning from a Causal Perspective
Lin, Guoliang
Xu, Yongheng
Lai, Hanjiang
Yin, Jian
Machine Learning
Artificial Intelligence
Few-shot learning with $N$-way $K$-shot scheme is an open challenge in machine learning. Many metric-based approaches have been proposed to tackle this problem, e.g., the Matching Networks and CLIP-Adapter. Despite that these approaches have shown significant progress, the mechanism of why these methods succeed has not been well explored. In this paper, we try to interpret these metric-based few-shot learning methods via causal mechanism. We show that the existing approaches can be viewed as specific forms of front-door adjustment, which can alleviate the effect of spurious correlations and thus learn the causality. This causal interpretation could provide us a new perspective to better understand these existing metric-based methods. Further, based on this causal interpretation, we simply introduce two causal methods for metric-based few-shot learning, which considers not only the relationship between examples but also the diversity of representations. Experimental results demonstrate the superiority of our proposed methods in few-shot classification on various benchmark datasets. Code is available in https://github.com/lingl1024/causalFewShot.
title Revisiting Few-Shot Learning from a Causal Perspective
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2209.13816