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 unbiased models on imbalanced datasets is a significant challenge.Rare classes tend to get a concentrated representation in the classificationspace which hampers the generalization of learned boundaries to new testexamples. In this paper, we demonstrate that the Bayesian uncertainty estimatesdirectly correlate with the rarity of classes and the difficulty level ofindividual samples. Subsequently, we present a novel framework for uncertaintybased class imbalance learning that follows two key insights: First,classification boundaries should be extended further away from a more uncertain(rare) class to avoid overfitting and enhance its generalization. Second, eachsample should be modeled as a multi-variate Gaussian distribution with a meanvector and a covariance matrix defined by the sample's uncertainty. The learnedboundaries should respect not only the individual samples but also theirdistribution in the feature space. Our proposed approach efficiently utilizessample and class uncertainty information to learn robust features and moregeneralizable classifiers. We systematically study the class imbalance problemand derive a novel loss formulation for max-margin learning based on Bayesianuncertainty measure. The proposed method shows significant performanceimprovements on six benchmark datasets for face verification, attributeprediction, digit/object classification and skin lesion detection.

Striking the Right Balance with Uncertainty