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

In computer vision, the performance of deep neural networks (DNNs) is highlyrelated to the feature extraction ability, i.e., the ability to recognize andfocus on key pixel regions in an image. However, in this paper, wequantitatively and statistically illustrate that DNNs have a serious attentionbias problem on many samples from some popular datasets: (1) Position bias:DNNs fully focus on label-independent regions; (2) Range bias: The focusedregions from DNN are not completely contained in the ideal region. Moreover, wefind that the existing self-attention modules can alleviate these biases to acertain extent, but the biases are still non-negligible. To further mitigatethem, we propose a lightweight sub-attention strategy (LSAS), which utilizeshigh-order sub-attention modules to improve the original self-attentionmodules. The effectiveness of LSAS is demonstrated by extensive experiments onwidely-used benchmark datasets and popular attention networks. We release ourcode to help other researchers to reproduce the results ofLSAS~\footnote{https://github.com/Qrange-group/LSAS}.

LSAS: Lightweight Sub-attention Strategy for Alleviating Attention Bias Problem