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

Supporting model interpretability for complex phenomena where annotators canlegitimately disagree, such as emotion recognition, is a challenging machinelearning task. In this work, we show that explicitly quantifying theuncertainty in such settings has interpretability benefits. We use a simplemodification of a classical network inference using Monte Carlo dropout to givemeasures of epistemic and aleatoric uncertainty. We identify a significantcorrelation between aleatoric uncertainty and human annotator disagreement($r\approx.3$). Additionally, we demonstrate how difficult and subjectivetraining samples can be identified using aleatoric uncertainty and howepistemic uncertainty can reveal data bias that could result in unfairpredictions. We identify the total uncertainty as a suitable surrogate formodel calibration, i.e. the degree we can trust model's predicted confidence.In addition to explainability benefits, we observe modest performance boostsfrom incorporating model uncertainty.

Characterizing Sources of Uncertainty to Proxy Calibration and Disambiguate Annotator and Data Bias