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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

In recent years, many incidents have been reported where machine learningmodels exhibited discrimination among people based on race, sex, age, etc.Research has been conducted to measure and mitigate unfairness in machinelearning models. For a machine learning task, it is a common practice to builda pipeline that includes an ordered set of data preprocessing stages followedby a classifier. However, most of the research on fairness has considered asingle classifier based prediction task. What are the fairness impacts of thepreprocessing stages in machine learning pipeline? Furthermore, studies showedthat often the root cause of unfairness is ingrained in the data itself, ratherthan the model. But no research has been conducted to measure the unfairnesscaused by a specific transformation made in the data preprocessing stage. Inthis paper, we introduced the causal method of fairness to reason about thefairness impact of data preprocessing stages in ML pipeline. We leveragedexisting metrics to define the fairness measures of the stages. Then weconducted a detailed fairness evaluation of the preprocessing stages in 37pipelines collected from three different sources. Our results show that certaindata transformers are causing the model to exhibit unfairness. We identified anumber of fairness patterns in several categories of data transformers.Finally, we showed how the local fairness of a preprocessing stage composes inthe global fairness of the pipeline. We used the fairness composition to chooseappropriate downstream transformer that mitigates unfairness in the machinelearning pipeline.

Fair Preprocessing: Towards Understanding Compositional Fairness of Data Transformers in Machine Learning Pipeline