# Revisiting the Transferability of Few-Shot Image Classification: A Frequency Spectrum Perspective

## TL;DR

This paper explores how different frequency components in images affect few-shot image classification performance and introduces a new method to improve transferability.

## Contribution

The paper introduces a novel frequency spectrum perspective and a method called Frequency Spectrum Mask (FRSM) to enhance transferability in few-shot image classification.

## Key findings

- Non-causal frequencies like background information negatively impact transferability in few-shot learning.
- The proposed FRSM method improves transferability by weighting and mitigating non-causal frequencies.
- Experiments show FRSM significantly enhances performance across nine datasets.

## Abstract

Few-shot learning, especially few-shot image classification (FSIC), endeavors to recognize new categories using only a handful of labeled images by transferring knowledge from a model trained on base categories. Despite numerous efforts to address the challenge of deficient transferability caused by the distribution shift between the base and new classes, the fundamental principles remain a subject of debate. In this paper, we elucidate why a decline in performance occurs and what information is transferred during the testing phase, examining it from a frequency spectrum perspective. Specifically, we adopt causality on the frequency space for FSIC. With our causal assumption, non-causal frequencies (e.g., background knowledge) act as confounders between causal frequencies (e.g., object information) and predictions. Our experimental results reveal that different frequency components represent distinct semantics, and non-causal frequencies adversely affect transferability, resulting in suboptimal performance. Subsequently, we suggest a straightforward but potent approach, namely the Frequency Spectrum Mask (FRSM), to weight the frequency and mitigate the impact of non-causal frequencies. Extensive experiments demonstrate that the proposed FRSM method significantly enhanced the transferability of the FSIC model across nine testing datasets.