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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Despite the growing demand for tuning foundation vision transformers (FViTs)on downstream tasks, fully unleashing FViTs' potential under data-limitedscenarios (e.g., few-shot tuning) remains a challenge due to FViTs' data-hungrynature. Common data augmentation techniques fall short in this context due tothe limited features contained in the few-shot tuning data. To tackle thischallenge, we first identify an opportunity for FViTs in few-shot tuning:pretrained FViTs themselves have already learned highly representative featuresfrom large-scale pretraining data, which are fully preserved during widely usedparameter-efficient tuning. We thus hypothesize that leveraging those learnedfeatures to augment the tuning data can boost the effectiveness of few-shotFViT tuning. To this end, we propose a framework called Hint-based DataAugmentation (Hint-Aug), which aims to boost FViT in few-shot tuning byaugmenting the over-fitted parts of tuning samples with the learned features ofpretrained FViTs. Specifically, Hint-Aug integrates two key enablers: (1) anAttentive Over-fitting Detector (AOD) to detect over-confident patches offoundation ViTs for potentially alleviating their over-fitting on the few-shottuning data and (2) a Confusion-based Feature Infusion (CFI) module to infuseeasy-to-confuse features from the pretrained FViTs with the over-confidentpatches detected by the above AOD in order to enhance the feature diversityduring tuning. Extensive experiments and ablation studies on five datasets andthree parameter-efficient tuning techniques consistently validate Hint-Aug'seffectiveness: 0.04% ~ 32.91% higher accuracy over the state-of-the-art (SOTA)data augmentation method under various low-shot settings. For example, on thePet dataset, Hint-Aug achieves a 2.22% higher accuracy with 50% less trainingdata over SOTA data augmentation methods.

Hint-Aug: Drawing Hints from Foundation Vision Transformers Towards Boosted Few-Shot Parameter-Efficient Tuning