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

Machine learning model bias can arise from dataset composition: sensitivefeatures correlated to the learning target disturb the model decision rule andlead to performance differences along the features. Existing de-biasing workcaptures prominent and delicate image features which are traceable in modellatent space, like colors of digits or background of animals. However, usingthe latent space is not sufficient to understand all dataset featurecorrelations. In this work, we propose a framework to extract feature clustersin a dataset based on image descriptions, allowing us to capture both subtleand coarse features of the images. The feature co-occurrence pattern isformulated and correlation is measured, utilizing a human-in-the-loop forexamination. The analyzed features and correlations are human-interpretable, sowe name the method Common-Sense Bias Discovery (CSBD). Having exposed sensitivecorrelations in a dataset, we demonstrate that downstream model bias can bemitigated by adjusting image sampling weights, without requiring a sensitivegroup label supervision. Experiments show that our method discovers novelbiases on multiple classification tasks for two benchmark image datasets, andthe intervention outperforms state-of-the-art unsupervised bias mitigationmethods.

Common-Sense Bias Discovery and Mitigation for Classification Tasks