A Confidence-Based Approach for Balancing Fairness and Accuracy. Fish, B., Kun, J., & Lelkes, ?. In Proceedings of the 2016 SIAM International Conference on Data Mining, of Proceedings, pages 144--152. Society for Industrial and Applied Mathematics, June, 2016.
A Confidence-Based Approach for Balancing Fairness and Accuracy [link]Paper  abstract   bibtex   
We study three classical machine learning algorithms in the context of algorithmic fairness: adaptive boosting, support vector machines, and logistic regression. Our goal is to maintain the high accuracy of these learning algorithms while reducing the degree to which they discriminate against individuals because of their membership in a protected group. Our first contribution is a method for achieving fairness by shifting the decision boundary for the protected group. The method is based on the theory of margins for boosting. Our method performs comparably to or outperforms previous algorithms in the fairness literature in terms of accuracy and low discrimination, while simultaneously allowing for a fast and transparent quantification of the trade-off between bias and error. Our second contribution addresses the shortcomings of the bias-error trade-off studied in most of the algorithmic fairness literature. We demonstrate that even hopelessly naive modifications of a biased algorithm, which cannot be reasonably said to be fair, can still achieve low bias and high accuracy. To help to distinguish between these naive algorithms and more sensible algorithms we propose a new measure of fairness, called resilience to random bias (RRB). We demonstrate that RRB distinguishes well between our naive and sensible fairness algorithms. RRB together with bias and accuracy provides a more complete picture of the fairness of an algorithm.
@incollection{fish_confidence-based_2016,
	series = {Proceedings},
	title = {A {Confidence}-{Based} {Approach} for {Balancing} {Fairness} and {Accuracy}},
	url = {http://epubs.siam.org/doi/10.1137/1.9781611974348.17},
	abstract = {We study three classical machine learning algorithms in the context of algorithmic fairness: adaptive boosting, support vector machines, and logistic regression. Our goal is to maintain the high accuracy of these learning algorithms while reducing the degree to which they discriminate against individuals because of their membership in a protected group. Our first contribution is a method for achieving fairness by shifting the decision boundary for the protected group. The method is based on the theory of margins for boosting. Our method performs comparably to or outperforms previous algorithms in the fairness literature in terms of accuracy and low discrimination, while simultaneously allowing for a fast and transparent quantification of the trade-off between bias and error. Our second contribution addresses the shortcomings of the bias-error trade-off studied in most of the algorithmic fairness literature. We demonstrate that even hopelessly naive modifications of a biased algorithm, which cannot be reasonably said to be fair, can still achieve low bias and high accuracy. To help to distinguish between these naive algorithms and more sensible algorithms we propose a new measure of fairness, called resilience to random bias (RRB). We demonstrate that RRB distinguishes well between our naive and sensible fairness algorithms. RRB together with bias and accuracy provides a more complete picture of the fairness of an algorithm.},
	urldate = {2016-12-13},
	booktitle = {Proceedings of the 2016 {SIAM} {International} {Conference} on {Data} {Mining}},
	publisher = {Society for Industrial and Applied Mathematics},
	author = {Fish, B. and Kun, J. and Lelkes, ?.},
	month = jun,
	year = {2016},
	keywords = {machinelearning, fairness},
	pages = {144--152},
	annote = {The authors evaluate three ML algorithms in terms of fairness, developing a resilience to random bias (RRB) measure and a shifted decision boundary (SDB) method in the process. The SDB method makes visible sources of bias. The three ML algorithms they evaluate are support vector machine (SVM), logistic regression, and Adaboost. They test their measures and methods on the 1994 Census Income dataset and German credit datasets, and conclude that their RRB measure is a "novel approach".},
	file = {Full Text PDF:C\:\\Users\\Ashudeep Singh\\Zotero\\storage\\Z2GAXS8C\\Fish et al. - 2016 - A Confidence-Based Approach for Balancing Fairness.pdf:application/pdf;Snapshot:C\:\\Users\\Ashudeep Singh\\Zotero\\storage\\96KFWM84\\1.9781611974348.html:text/html}
}

Downloads: 0