Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

FairGLVQ: Fairness in Partition-Based Classification

Felix Störck, Fabian Hinder, Johannes Brinkrolf, Benjamin Paassen, Valerie Vaquet, Barbara Hammer

TL;DR

This work develops a general framework for fair machine learning of partition-based models that does not depend on a specific fairness definition, and derives a fair version of learning vector quantization (LVQ) as a specific instantiation.

Abstract

Fairness is an important objective throughout society. From the distribution of limited goods such as education, over hiring and payment, to taxes, legislation, and jurisprudence. Due to the increasing importance of machine learning approaches in all areas of daily life including those related to health, security, and equity, an increasing amount of research focuses on fair machine learning. In this work, we focus on the fairness of partition- and prototype-based models. The contribution of this work is twofold: 1) we develop a general framework for fair machine learning of partition-based models that does not depend on a specific fairness definition, and 2) we derive a fair version of learning vector quantization (LVQ) as a specific instantiation. We compare the resulting algorithm against other algorithms from the literature on theoretical and real-world data showing its practical relevance.

FairGLVQ: Fairness in Partition-Based Classification

TL;DR

This work develops a general framework for fair machine learning of partition-based models that does not depend on a specific fairness definition, and derives a fair version of learning vector quantization (LVQ) as a specific instantiation.

Abstract

Fairness is an important objective throughout society. From the distribution of limited goods such as education, over hiring and payment, to taxes, legislation, and jurisprudence. Due to the increasing importance of machine learning approaches in all areas of daily life including those related to health, security, and equity, an increasing amount of research focuses on fair machine learning. In this work, we focus on the fairness of partition- and prototype-based models. The contribution of this work is twofold: 1) we develop a general framework for fair machine learning of partition-based models that does not depend on a specific fairness definition, and 2) we derive a fair version of learning vector quantization (LVQ) as a specific instantiation. We compare the resulting algorithm against other algorithms from the literature on theoretical and real-world data showing its practical relevance.

Paper Structure

This paper contains 13 sections, 5 equations, 2 figures, 1 table, 1 algorithm.

Table of Contents

  1. Introduction
  2. Problem Setup
  3. Fair Machine Learning
  4. Learning Vector Quantization Models
  5. Related Work
  6. Method
  7. Fairness for General Partition-Based Models
  8. Fair Hebbian Learning, Pseudo-Classes, and FairGLVQ
  9. Algorithm Design
  10. Experiments
  11. Synthetic Data
  12. Real-world Benchmarks
  13. Conclusion and Discussion

Figures (2)

  • Figure 1: Synthetic datasets (a: "local" with linear model (black), b: "XOR", c: fair projection of "local"). Label is color/hatched; Protected attribute is shape.
  • Figure 2: Empirical evaluation of real-world datasets for different regularizations using 5-fold cross-validation. Fairness score ($x$-axis, lower is better) and Accuracy ($y$-axis, higher is better). COMPAS (top) and Adult (bottom).