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

Motivated by the growing importance of reducing unfairness in ML predictions,Fair-ML researchers have presented an extensive suite of algorithmic'fairness-enhancing' remedies. Most existing algorithms, however, are agnosticto the sources of the observed unfairness. As a result, the literaturecurrently lacks guiding frameworks to specify conditions under which eachalgorithmic intervention can potentially alleviate the underpinning cause ofunfairness. To close this gap, we scrutinize the underlying biases (e.g., inthe training data or design choices) that cause observational unfairness. Wepresent the conceptual idea and a first implementation of a bias-injectionsandbox tool to investigate fairness consequences of various biases and assessthe effectiveness of algorithmic remedies in the presence of specific types ofbias. We call this process the bias(stress)-testing of algorithmicinterventions. Unlike existing toolkits, ours provides a controlled environmentto counterfactually inject biases in the ML pipeline. This stylized setupoffers the distinct capability of testing fairness interventions beyondobservational data and against an unbiased benchmark. In particular, we cantest whether a given remedy can alleviate the injected bias by comparing thepredictions resulting after the intervention in the biased setting with truelabels in the unbiased regime-that is, before any bias injection. We illustratethe utility of our toolkit via a proof-of-concept case study on synthetic data.Our empirical analysis showcases the type of insights that can be obtainedthrough our simulations.

A Sandbox Tool to Bias(Stress)-Test Fairness Algorithms