Dataset Information

Learning Optimal Fair Policies.

ABSTRACT: Systematic discriminatory biases present in our society influence the way data is collected and stored, the way variables are defined, and the way scientific findings are put into practice as policy. Automated decision procedures and learning algorithms applied to such data may serve to perpetuate existing injustice or unfairness in our society. In this paper, we consider how to make optimal but fair decisions, which "break the cycle of injustice" by correcting for the unfair dependence of both decisions and outcomes on sensitive features (e.g., variables that correspond to gender, race, disability, or other protected attributes). We use methods from causal inference and constrained optimization to learn optimal policies in a way that addresses multiple potential biases which afflict data analysis in sensitive contexts, extending the approach of Nabi & Shpitser (2018). Our proposal comes equipped with the theoretical guarantee that the chosen fair policy will induce a joint distribution for new instances that satisfies given fairness constraints. We illustrate our approach with both synthetic data and real criminal justice data.

SUBMITTER: Nabi R

PROVIDER: S-EPMC6935348 | biostudies-literature | 2019 Jun

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Learning Optimal Fair Policies.

Nabi Razieh R Malinsky Daniel D Shpitser Ilya I

Proceedings of machine learning research 20190601

Systematic discriminatory biases present in our society influence the way data is collected and stored, the way variables are defined, and the way scientific findings are put into practice as policy. Automated decision procedures and learning algorithms applied to such data may serve to perpetuate existing injustice or unfairness in our society. In this paper, we consider how to make optimal but fair decisions, which "break the cycle of injustice" by correcting for the unfair dependence of both ...[more]

PMID: 31886463

Similar Datasets

Project description:BackgroundIn Europe, progress in the development of health policies that address the needs of migrants and ethnic minorities has been slow. This is partly due to the absence of a strategic commitment by the health authorities. The Ministry of Public Health commissioned the ETHEALTH (EThnicity &HEALTH) group to formulate relevant recommendations to the public authorities with a view to reducing health inequalities among ethnic minorities. This paper describes the political process and the outcomes of the ETHEALTH expert group.ResultsAfter ten meetings, the ETHEALTH group came up with 46 recommendations, which were presented at a national press conference in December 2011. Target groups concerned by these recommendations covered both irregular migrants and migrants entitled to the national insurance coverage. Recommendations were supported by the need of combining universal approaches to health care with more specific approaches. The scope of the recommendations concerned health care as well as prevention, health promotion and access to health care. When analysing the content of the recommendations, some ETHEALTH recommendations were not fully measurable, and time-related; they were, however, quite specific and realistic within the Belgian context. The weak political commitment of an executive agency was identified as a major obstacle to the implementation of the recommendations.ConclusionsThe ETHEALTH group was an example of scientific advice on a global health issue. It also demonstrated the feasibility of coming up with a comprehensive strategy to decrease ethnic health inequalities, even in a political context where migration issues are sensitive. Two final lessons may be highlighted at the end of the first phase of the ETHEALTH project: firstly, the combination of scientific knowledge and practical expertise makes recommendations SMART; and, secondly, the low level of commitment on the part of policymakers might jeopardise the effective implementation of the recommendations.

Dataset Information

Learning Optimal Fair Policies.

Publications

Learning Optimal Fair Policies.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure