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

Dense prediction models are widely used for image segmentation. One importantchallenge is to sufficiently train these models to yield good generalizationsfor hard-to-learn pixels. A typical group of such hard-to-learn pixels areboundaries between instances. Many studies have proposed to give specificattention to learning the boundary pixels. They include designing multi-tasknetworks with an additional task of boundary prediction and increasing theweights of boundary pixels' predictions in the loss function. Such strategiesrequire defining what to attend beforehand and incorporating this definedattention to the learning model. However, there may exist other groups ofhard-to-learn pixels and manually defining and incorporating the appropriateattention for each group may not be feasible. In order to provide a moreattainable and scalable solution, this paper proposes AttentionBoost, which isa new multi-attention learning model based on adaptive boosting. AttentionBoostdesigns a multi-stage network and introduces a new loss adjustment mechanismfor a dense prediction model to adaptively learn what to attend at each stagedirectly on image data without necessitating any prior definition about what toattend. This mechanism modulates the attention of each stage to correct themistakes of previous stages, by adjusting the loss weight of each pixelprediction separately with respect to how accurate the previous stages are onthis pixel. This mechanism enables AttentionBoost to learn different attentionsfor different pixels at the same stage, according to difficulty of learningthese pixels, as well as multiple attentions for the same pixel at differentstages, according to confidence of these stages on their predictions for thispixel. Using gland segmentation as a showcase application, our experimentsdemonstrate that AttentionBoost improves the results of its counterparts.

AttentionBoost: Learning What to Attend by Boosting Fully Convolutional Networks