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

Traditional deep learning (DL) suffers from two core problems. Firstly, itassumes training samples are independent and identically distributed. However,numerous real-world datasets group samples by shared measurements (e.g., studyparticipants or cells), violating this assumption. In these scenarios, DL canshow compromised performance, limited generalization, and interpretabilityissues, coupled with cluster confounding causing Type 1 and 2 errors. Secondly,models are typically trained for overall accuracy, often neglectingunderrepresented groups and introducing biases in crucial areas like loanapprovals or determining health insurance rates, such biases can significantlyimpact one's quality of life. To address both of these challengessimultaneously, we present a mixed effects deep learning (MEDL) framework. MEDLseparately quantifies cluster-invariant fixed effects (FE) and cluster-specificrandom effects (RE) through the introduction of: 1) a cluster adversary whichencourages the learning of cluster-invariant FE, 2) a Bayesian neural networkwhich quantifies the RE, and a mixing function combining the FE an RE into amixed-effect prediction. We marry this MEDL with adversarial debiasing, whichpromotes equality-of-odds fairness across FE, RE, and ME predictions forfairness-sensitive variables. We evaluated our approach using three datasets:two from census/finance focusing on income classification and one fromhealthcare predicting hospitalization duration, a regression task. Ourframework notably enhances fairness across all sensitive variables-increasingfairness up to 82% for age, 43% for race, 86% for sex, and 27% formarital-status. Besides promoting fairness, our method maintains the robustperformance and clarity of MEDL. It's versatile, suitable for various datasettypes and tasks, making it broadly applicable. Our GitHub repository houses theimplementation.

Fairness-enhancing mixed effects deep learning improves fairness on in- and out-of-distribution clustered (non-iid) data