PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
April 19, 2026IET Computer Vision0 citationsOpen Access

ESFFA: Early‐Stage Feature Frequency Attack in Cross‐Domain Few‐Shot Learning

View Full Paper
XWXu WangZCZiyang CaoJZJia Zhao

Key Points

  • The research focuses on improving cross-domain few-shot learning by addressing domain discrepancies through frequency-based augmentation techniques.
  • Proposed Early-Stage Feature Frequency Attack (ESFFA) to pertain low-frequency components along gradient directions.
  • Randomly masked high-frequency components in shallow-layer feature maps.
  • Conducted experiments using eight target datasets to validate the proposed method's effectiveness.
  • Demonstrated improved model generalization across various target datasets.
  • Effectively disrupted model reliance on simplistic frequency patterns.
  • Showed that frequency-space operations enhance performance compared to existing methods.

Abstract

ABSTRACT Cross‐domain few‐shot learning (CD‐FSL) addresses the challenge of few‐shot classification under significant distribution shifts between source and novel target domains. The core difficulty lies in bridging the domain gap. Existing methods primarily mitigate this issue from the spatial perspective, overlooking the role of frequency information. Empirical studies reveal that augmenting samples through frequency‐space operations can alleviate domain discrepancies. However, current frequency‐based augmentation methods typically perform a holistic replacement of high‐frequency components, which oversimplifies the process and fails to adequately model complex frequency shortcuts (i.e., the tendency of models to prioritise learning the simplest and most class‐discriminative frequency patterns rather than semantically meaningful features). Inspired by gradient‐based adversarial learning, we propose Early‐Stage Feature Frequency Attack (ESFFA). Our method perturbs low‐frequency components along gradient directions and randomly masks high‐frequency components in shallow‐layer feature maps. This joint operation in the feature space, compared to directly replacing components in raw pixels, more effectively disrupts the model's reliance on frequency shortcuts. This approach compels the model to adapt to dynamically changing frequency characteristics, thereby enhancing cross‐domain generalisation. Experiments on eight target datasets validate its effectiveness.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Wang et al. (2026) studied this question.

synapsesocial.com/papers/69e472a8010ef96374d8e9ebhttps://doi.org/10.1049/cvi2.70065
Ask AI
Helpful
Bookmark
Share
View Full Paper