English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Learning with little data is challenging but often inevitable in variousapplication scenarios where the labeled data is limited and costly. Recently,few-shot learning (FSL) gained increasing attention because of itsgeneralizability of prior knowledge to new tasks that contain only a fewsamples. However, for data-intensive models such as vision transformer (ViT),current fine-tuning based FSL approaches are inefficient in knowledgegeneralization and thus degenerate the downstream task performances. In thispaper, we propose a novel mask-guided vision transformer (MG-ViT) to achieve aneffective and efficient FSL on ViT model. The key idea is to apply a mask onimage patches to screen out the task-irrelevant ones and to guide the ViT tofocus on task-relevant and discriminative patches during FSL. Particularly,MG-ViT only introduces an additional mask operation and a residual connection,enabling the inheritance of parameters from pre-trained ViT without any othercost. To optimally select representative few-shot samples, we also include anactive learning based sample selection method to further improve thegeneralizability of MG-ViT based FSL. We evaluate the proposed MG-ViT on bothAgri-ImageNet classification task and ACFR apple detection task withgradient-weighted class activation mapping (Grad-CAM) as the mask. Theexperimental results show that the MG-ViT model significantly improves theperformance when compared with general fine-tuning based ViT models, providingnovel insights and a concrete approach towards generalizing data-intensive andlarge-scale deep learning models for FSL.

Mask-guided Vision Transformer (MG-ViT) for Few-Shot Learning