FAIR: Fair Adversarial Instance Re-weighting
Andrija Petrovi\'c, Mladen Nikoli\'c, Sandro Radovanovi\'c, Boris, Deliba\v{s}i\'c, Milo\v{s} Jovanovi\'c

TL;DR
FAIR introduces a novel method combining adversarial training and instance re-weighting to improve fairness in machine learning, offering interpretability and better accuracy-unfairness trade-offs.
Contribution
It is the first model to merge reweighting and adversarial approaches with an interpretable weighting function for fairness.
Findings
Outperforms 7 state-of-the-art models in fairness-accuracy trade-off
Provides a fully probabilistic framework for fairness modeling
Offers four variants and extensive theoretical analysis
Abstract
With growing awareness of societal impact of artificial intelligence, fairness has become an important aspect of machine learning algorithms. The issue is that human biases towards certain groups of population, defined by sensitive features like race and gender, are introduced to the training data through data collection and labeling. Two important directions of fairness ensuring research have focused on (i) instance weighting in order to decrease the impact of more biased instances and (ii) adversarial training in order to construct data representations informative of the target variable, but uninformative of the sensitive attributes. In this paper we propose a Fair Adversarial Instance Re-weighting (FAIR) method, which uses adversarial training to learn instance weighting function that ensures fair predictions. Merging the two paradigms, it inherits desirable properties from both --…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
MethodsInterpretability
