Abstract: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 <a class="link-external link-https" href="https://github.com/lingl1024/causalFewShot" rel="external noopener nofollow">this https URL</a>.

What problem does this paper attempt to address?

The problems that this paper attempts to solve are as follows: In few - shot learning (FSL) tasks, although existing metric - based methods have made remarkable progress, their success mechanisms have not been fully explored. Moreover, due to the limited training data, spurious correlations are likely to occur, thus affecting the generalization ability of the model. Specifically, the paper focuses on the N - way K - shot classification task, that is, there are N classes and only K labeled training samples for each class. Under such a setting, the model is easily influenced by unobserved confounders, leading to learning spurious correlations instead of real causal relationships. For example, when the background of all bird pictures is the sky, the model may wrongly associate "sky" with "bird". To solve these problems, the paper attempts to re - interpret existing metric - based FSL methods from the perspective of causal inference and proposes two new causal methods to improve the performance of FSL tasks. By introducing causal mechanisms, especially front - door adjustment, the impact of spurious correlations can be reduced, so as to better learn the causal relationship between samples and labels. ### Main contributions of the paper: 1. **Formalized the causal framework in FSL**: The paper found that existing mainstream metric - based FSL methods can be well - explained in the causal framework. 2. **Proposed two causal methods**: These methods consider not only the interaction between samples but also the diversity of representations. 3. **Experimental verification**: Evaluated on multiple benchmark datasets and explored the characteristics of these methods through ablation studies. The experimental results show that the proposed methods have significant performance improvements. ### Summary: The paper aims to understand and improve few - shot learning tasks from the perspective of causal inference, especially for spurious correlation and data bias problems. By introducing causal mechanisms, the paper provides a new theoretical explanation and designs more effective FSL algorithms.

Revisiting Few-Shot Learning from a Causal Perspective

PatchMix Augmentation to Identify Causal Features in Few-shot Learning

Interventional Few-Shot Learning

A Hybrid Approach with Optimization-Based and Metric-Based Meta-Learner for Few-Shot Learning

Disentangle and Remerge: Interventional Knowledge Distillation for Few-Shot Object Detection from a Conditional Causal Perspective

Few-Shot Learning Based on Deep Learning for Image Classification

Reweighting and Information-Guidance Networks for Few-Shot Learning

Causal Interventions-based Few-Shot Named Entity Recognition

Knowledge-Based Fine-Grained Classification for Few-Shot Learning.

Category Relevance Redirection Network for Few-Shot Classification

Adaptive Cross-Modal Few-Shot Learning

Enhancing Few-Shot Classification without Forgetting through Multi-Level Contrastive Constraints

A Closer Look at Few-Shot Video Classification: A New Baseline and Benchmark

Less is More: A Closer Look at Semantic-based Few-Shot Learning

Few-shot Learning with Meta Metric Learners

Learning Causal Domain-Invariant Temporal Dynamics for Few-Shot Action Recognition

Distribution Consistency Based Covariance Metric Networks for Few-Shot Learning.

Transductive Episodic-Wise Adaptive Metric for Few-Shot Learning

Boosting Few-Shot Classification with View-Learnable Contrastive Learning

Exploring Lottery Ticket Hypothesis in Few-Shot Learning

The Devil is in the Few Shots: Iterative Visual Knowledge Completion for Few-shot Learning