Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

FAIREDU: A Multiple Regression-Based Method for Enhancing Fairness in Machine Learning Models for Educational Applications

Nga Pham, Minh Kha Do, Tran Vu Dai, Pham Ngoc Hung, Anh Nguyen-Duc

TL;DR

FAIREDU tackles fairness in educational AI/ML by addressing intersectionality across multiple sensitive features using a regression-based preprocessing strategy. It extends prior work LTDD with a multivariate regression framework to remove dependencies between non-sensitive features and several sensitive attributes, producing a bias-reduced dataset before model training. Across seven educational datasets and three common models, FAIREDU improves fairness under DI, SPD, AOD, and EOD metrics with only minor declines in accuracy and recall, outperforming several state-of-the-art baselines in most cases. The work demonstrates the method’s potential to reduce fairness gaps across multiple groups in educational settings, while acknowledging limitations related to linearity assumptions and generalizability, and outlines future directions for broader datasets, adaptive interventions, and new fairness metrics.

Abstract

Fairness in artificial intelligence and machine learning (AI/ML) models is becoming critically important, especially as decisions made by these systems impact diverse groups. In education, a vital sector for all countries, the widespread application of AI/ML systems raises specific concerns regarding fairness. Current research predominantly focuses on fairness for individual sensitive features, which limits the comprehensiveness of fairness assessments. This paper introduces FAIREDU, a novel and effective method designed to improve fairness across multiple sensitive features. Through extensive experiments, we evaluate FAIREDU effectiveness in enhancing fairness without compromising model performance. The results demonstrate that FAIREDU addresses intersectionality across features such as gender, race, age, and other sensitive features, outperforming state-of-the-art methods with minimal effect on model accuracy. The paper also explores potential future research directions to enhance further the method robustness and applicability to various machine-learning models and datasets.

FAIREDU: A Multiple Regression-Based Method for Enhancing Fairness in Machine Learning Models for Educational Applications

TL;DR

FAIREDU tackles fairness in educational AI/ML by addressing intersectionality across multiple sensitive features using a regression-based preprocessing strategy. It extends prior work LTDD with a multivariate regression framework to remove dependencies between non-sensitive features and several sensitive attributes, producing a bias-reduced dataset before model training. Across seven educational datasets and three common models, FAIREDU improves fairness under DI, SPD, AOD, and EOD metrics with only minor declines in accuracy and recall, outperforming several state-of-the-art baselines in most cases. The work demonstrates the method’s potential to reduce fairness gaps across multiple groups in educational settings, while acknowledging limitations related to linearity assumptions and generalizability, and outlines future directions for broader datasets, adaptive interventions, and new fairness metrics.

Abstract

Fairness in artificial intelligence and machine learning (AI/ML) models is becoming critically important, especially as decisions made by these systems impact diverse groups. In education, a vital sector for all countries, the widespread application of AI/ML systems raises specific concerns regarding fairness. Current research predominantly focuses on fairness for individual sensitive features, which limits the comprehensiveness of fairness assessments. This paper introduces FAIREDU, a novel and effective method designed to improve fairness across multiple sensitive features. Through extensive experiments, we evaluate FAIREDU effectiveness in enhancing fairness without compromising model performance. The results demonstrate that FAIREDU addresses intersectionality across features such as gender, race, age, and other sensitive features, outperforming state-of-the-art methods with minimal effect on model accuracy. The paper also explores potential future research directions to enhance further the method robustness and applicability to various machine-learning models and datasets.
Paper Structure (26 sections, 4 equations, 4 figures, 7 tables, 1 algorithm)

This paper contains 26 sections, 4 equations, 4 figures, 7 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Background
  3. Fairness for Machine Learning Systems
  4. Sensitive features
  5. Detecting and fixing fairness bugs for AI/ML systems
  6. FAIREDU - A regression-based method for fairness of multiple sensitive features in Education
  7. Idea development
  8. Overall architecture of FAIREDU
  9. Algorithm
  10. Experiment Setup
  11. Selection of Datasets
  12. Selection of models
  13. Evaluation Metrics
  14. Results
  15. RQ1 - Is there a systematic bias present among sensitive features within educational datasets?
  16. ...and 11 more sections

Figures (4)

  • Figure 1: The overall architecture of FAIREDU
  • Figure 2: Comparing $|1-DI|$ of sensitive features across datasets
  • Figure 3: Comparison of SPD across methods
  • Figure 4: Comparison of $|1-DI|$ across methods