Definition Of Fairness Research Articles

How do fairness definitions fare? Testing public attitudes towards three algorithmic definitions of fairness in loan allocations

What is the best way to define algorithmic fairness? While many definitions of fairness have been proposed in the computer science literature, there is no clear agreement over a particular definition. In this work, we investigate ordinary people's perceptions of three of these fairness definitions. Across three online experiments, we test which definitions people perceive to be the fairest in the context of loan decisions, and whether fairness perceptions change with the addition of sensitive information (i.e., race or gender of the loan applicants). Overall, one definition (calibrated fairness) tends to be more preferred than the others, and the results also provide support for the principle of affirmative action.

The Age of Secrecy and Unfairness in Recidivism Prediction

In our current society, secret algorithms make important decisions about individuals. There has been substantial discussion about whether these algorithms are unfair to groups of individuals. While noble, this pursuit is complex and ultimately stagnating because there is no clear definition of fairness and competing definitions are largely incompatible. We argue that the focus on the question of fairness is misplaced, as these algorithms fail to meet a more important and yet readily obtainable goal: transparency. As a result, creators of secret algorithms can provide incomplete or misleading descriptions about how their models work, and various other kinds of errors can easily go unnoticed. By trying to partially reconstruct the COMPAS model—a recidivism risk-scoring model used throughout the criminal justice system—we show that it does not seem to depend linearly on the defendant’s age, despite statements to the contrary by the model’s creator. This observation has not been made before despite many recently published papers on COMPAS. Furthermore, by subtracting from COMPAS its (hypothesized) nonlinear age component, we show that COMPAS does not necessarily depend on race other than through age and criminal history. This contradicts ProPublica’s analysis, which made assumptions about age that disagree with what we observe in the data. In other words, faulty assumptions about a proprietary model led to faulty conclusions that went unchecked until now. Were the model transparent in the first place, this likely would not have occurred. We demonstrate other issues with definitions of fairness and lack of transparency in the context of COMPAS, including that a simple model based entirely on a defendant’s age is as ‘unfair’ as COMPAS by ProPublica’s chosen definition. We find that there are many defendants with low risk scores but long criminal histories, suggesting that data inconsistencies occur frequently in criminal justice databases. We argue that transparency satisfies a different notion of procedural fairness by providing both the defendants and the public with the opportunity to scrutinize the methodology and calculations behind risk scores for recidivism.

Algorithmic Fairness in Mortgage Lending: From Absolute Conditions to Relational Trade-Offs

To address the rising concern that algorithmic decision-making may reinforce discriminatory biases, researchers have proposed many notions of fairness and corresponding mathematical formalizations. Each of these notions is often presented as a one-size-fits-all, absolute condition; however, in reality, the practical and ethical trade-offs are unavoidable and more complex. We introduce a new approach that considers fairness—not as a binary, absolute mathematical condition—but rather, as a relational notion in comparison to alternative decision-making processes. Using US mortgage lending as an example use case, we discuss the ethical foundations of each definition of fairness and demonstrate that our proposed methodology more closely captures the ethical trade-offs of the decision-maker, as well as forcing a more explicit representation of which values and objectives are prioritised.

Fairness in Measurement and Selection: Statistical, Philosophical, and Public Perspectives

AbstractSelection decisions have a major impact on our education, occupation, and quality of life, and the role of standardized tests in selection has always been a source of controversy. Here, I consider various definitions of fairness in measurement and selection—those emerging from within educational measurement and statistics, those from philosophy, and finally, those from the public. I use examples of public challenges to selection practices to illustrate the fact that technical and philosophical definitions of fairness do not align well with public concerns. I emphasize the importance of promoting awareness of existing standards, advocating for the fair use of testing and selection practices, and communicating in a candid and straightforward way when engaging with test takers and test users.

Matching code and law: achieving algorithmic fairness with optimal transport

Increasingly, discrimination by algorithms is perceived as a societal and legal problem. As a response, a number of criteria for implementing algorithmic fairness in machine learning have been developed in the literature. This paper proposes the continuous fairness algorithm $$(\hbox {CFA}\theta )$$ which enables a continuous interpolation between different fairness definitions. More specifically, we make three main contributions to the existing literature. First, our approach allows the decision maker to continuously vary between specific concepts of individual and group fairness. As a consequence, the algorithm enables the decision maker to adopt intermediate “worldviews” on the degree of discrimination encoded in algorithmic processes, adding nuance to the extreme cases of “we’re all equal” and “what you see is what you get” proposed so far in the literature. Second, we use optimal transport theory, and specifically the concept of the barycenter, to maximize decision maker utility under the chosen fairness constraints. Third, the algorithm is able to handle cases of intersectionality, i.e., of multi-dimensional discrimination of certain groups on grounds of several criteria. We discuss three main examples (credit applications; college admissions; insurance contracts) and map out the legal and policy implications of our approach. The explicit formalization of the trade-off between individual and group fairness allows this post-processing approach to be tailored to different situational contexts in which one or the other fairness criterion may take precedence. Finally, we evaluate our model experimentally.

WE-Rec: A fairness-aware reciprocal recommendation based on Walrasian equilibrium

The emergence of online dating and recruiting platforms brings big challenges to the reciprocal recommendation which has attracted a lot of research attention. Most previous approaches improved the accuracy and diversity of reciprocal recommendations, but few researcher made efforts on the fairness-aware recommendation which aims to avoid the discrimination and mistreatment of vulnerable groups. In this paper, we concentrate on the research of fairness-aware recommendations in the reciprocal recommender system and propose an approach to rerank the recommendation list by optimizing three significant fairness-aware criteria between parties (i.e., buyers and sellers) based on Walrasian equilibrium: (1) the disparity of service; (2) the similarity of mutual preference; (3) the equilibrium of demand and supply. According to these definitions of fairness, we cast the task of reciprocal recommendation as a multi-objective optimization considering the satisfaction of individuals, the fairness of recommendations, and the market clearing simultaneously. The extensive experiments are conducted on two real-world datasets, and the results demonstrate the effectiveness of our approach.

Fairness or benefit? The effect of power on distributive fairness

<p id="C2">Distributive fairness is a basic behavioral norm and an important pursuit in our daily life, and it plays an important role in our social interactions. In studies of distributive fairness, individuals' fairness perception and the factors affecting it have gathered much research attention. Previous research shows that the degree of fairness allocation can affect individuals' fairness perception. Based on the equity theory of fairness, equal allocation can be perceived as fairness for individuals if they have equal ability and equal contribution to the allocation; less than equal allocation may be perceived as disadvantageous inequality, and more than equal allocation as advantageous inequality. Previous studies also suggest that social situations, such as social hierarchy and social distance, can affect individuals' fairness perception. In real life, resource allocation often involves power situations, in which individuals may have different levels of power. How power influence individuals' fairness perception? Is the fairness perception power-dependent? So far, few studies have explored the effect of power on fairness perception directly. The present study aims to address this question through three experiments. <br/><p id="C70">Based on the approach-inhibition theory, the powerful usually expect themselves to be surrounded by rewards and lack of threats, while the powerless usually expect themselves to be surrounded by threats and lack of rewards. Previous research shows that individuals' cognition and behavior can be affected by their internal expectations, and they are often sensitive to outcomes that violate their own expectations. Equal allocation and advantageous inequality allocation mean reward, while disadvantageous inequality means threat. Thus, equal allocation and advantageous inequality allocation are the expectation of the powerful, and disadvantageous inequality allocation is the expectation of the powerless. Therefore, we hypothesize that the interaction between power and fairness degree can impact individuals' fairness perception. <br/><p id="C71">Three experiments were designed to test the hypothesis. To provide objective basis for the definitions of fairness, disadvantage inequality, and advantage inequality, we measured individuals' fairness ratings on different allocations in both the powerful and the powerless situations, after situation priming that manipulated power perception, in Experiment 1. In Experiment 2, we further investigated individuals' fairness ratings on three different allocations (i.e., disadvantageous unequal allocation, equal allocation, advantageous unequal allocation) in both the powerful and the powerless situations. We also recorded individuals' reaction times of fairness rating in Experiment 2 to gather extra evidence for fairness perception. In Experiment 3, a different manipulation method, role playing, was used to prime power. Same as in Experiment 2, we also recorded the fairness ratings and reaction times of the powerful and the powerless in Experiment 3. <br/><p id="C72">In terms of fairness rating scores, the powerful rated the fairness degree of equal allocation and advantageous unequal allocation higher than the powerless, and they rated the fairness degree of disadvantageous unequal allocation lower than the powerless. In reaction time, the powerful reacted faster than the powerless no matter what the allocation was. These results suggested that individuals' distributive fairness perception is power-dependent, supporting our hypothesis. The present findings provide experimental evidence for the approach-inhibition theory of power and the equity theory of fairness. They also improve our understanding of the relationship between power and fairness perception.

Bayesian Fairness

We consider the problem of how decision making can be fair when the underlying probabilistic model of the world is not known with certainty. We argue that recent notions of fairness in machine learning need to explicitly incorporate parameter uncertainty, hence we introduce the notion of Bayesian fairness as a suitable candidate for fair decision rules. Using balance, a definition of fairness introduced in (Kleinberg, Mullainathan, and Raghavan 2016), we show how a Bayesian perspective can lead to well-performing and fair decision rules even under high uncertainty.

A New Definition of Fairness for Non-Orthogonal Multiple Access

Fairness in non-orthogonal multiple access (NOMA) may be defined in different ways, taking into account the physical layer alone or also including the process of radio resource management. The latter has been frequently used in the recent literature, implying a significant advantage of this technique compared with conventional orthogonal multiple access (OMA). In this letter, we look at the fairness issue from another angle and introduce a fundamental definition of fairness, which measures the difference between the rates that can be achieved by the users and the fair rates suggested by the power distribution among them. With this definition, the fairness issue is inexistent in OMA and it appears as a NOMA-specific problem due to interference between users and the detection process. The new fairness index we advocate incorporates the power distribution of different users. Thus, in a cell with a non-uniform power distribution, fairness implies that a user with a higher power should get a higher rate, and hence, the index should measure the rate of each user by accounting for the fraction of total power allocated to it. Our index achieves its maximum value of 1 only when all users get their fair rates, and it goes to zero when one user gets all resources in the case of uniform power distribution. We provide an asymptotic analysis of fairness at high and low signal-to-noise ratio (SNR) values and give a comprehensive illustration in the two-user case.

Conceptualising fairness in classroom assessment: exploring the value of organisational justice theory

ABSTRACTFairness has recently moved into the spotlight as a core foundation of classroom assessment (CA). However, despite its significance for high-quality CA, fairness definitions and theories have been limited in the literature. Driven by the critiques directed at the ‘inadequacy’ and ‘fuzziness’ around CA fairness and recommendations to conceptualise fairness particularly for CA contexts, this paper aims to provide an explicit definition of CA fairness. Specifically, this paper brings together current scholarship in organisational justice theory and recent findings from the CA fairness literature to offer a more thorough conceptualisation. This conceptualisation not only presents a distinction between justice and fairness, but also provides a novel discussion of the relationship between justice and fairness with consideration for potential effects on students’ learning. The paper concludes with an agenda for further research on CA fairness.

Blanchard, A. and Strand, R. (eds) Cancer Biomarkers: Ethics, Economics and Society, Kokstad: Megaloceros Press2017. 149pp ISBN 978‐82‐91851‐04‐4

Blanchard, A. and Strand, R. (eds) Cancer Biomarkers: Ethics, Economics and Society, Kokstad: Megaloceros Press2017. 149pp ISBN 978‐82‐91851‐04‐4

Fairness Meets Machine Learning: Searching For A Better Balance

Machine learning (ML) affects nearly every aspect of our lives, including the weightiest ones such as criminal justice. As it becomes more widespread, however, it raises the question of how we can integrate fairness into ML algorithms to ensure that all citizens receive equal treatment and to avoid imperiling society’s democratic values. In this paper we study various formal definitions of fairness that can be embedded into ML algorithms and show that the root cause of most debates about AI fairness is society’s lack of a consistent understanding of fairness generally. We conclude that AI regulations stipulating an abstract fairness principle are ineffective societally. Capitalizing on extensive related work in computer science and the humanities, we present an approach that can help ML developers choose a formal definition of fairness suitable for a particular country and application domain. Abstract rules from the human world fail in the ML world and ML developers will never be free from criticism if the status quo remains. We argue that the law should shift from an abstract definition of fairness to a formal legal definition. Legislators and society as a whole should tackle the challenge of defining fairness, but since no definition perfectly matches the human sense of fairness, legislators must publicly acknowledge the drawbacks of the chosen definition and assert that the benefits outweigh them. Doing so creates transparent standards of fairness to ensure that technology serves the values and best interests of society.

Matching Code and Law: Achieving Algorithmic Fairness with Optimal Transport

Increasingly, discrimination by algorithms is perceived as a societal and legal problem. As a response, a number of criteria for implementing algorithmic fairness in machine learning have been developed in the literature. This paper proposes the Continuous Fairness Algorithm (CFAq) which enables a continuous interpolation between different fairness definitions. More specifically, we make three main contributions to the existing literature. First, our approach allows the decision maker to continuously vary between specific concepts of individual and group fairness. As a consequence, the algorithm enables the decision maker to adopt intermediate “worldviews” on the degree of discrimination encoded in algorithmic processes, adding nuance to the extreme cases of “we’re all equal” (WAE) and “what you see is what you get” (WYSIWYG) proposed so far in the literature. Second, we use optimal transport theory, and specifically the concept of the barycenter, to maximize decision maker utility under the chosen fairness constraints. Third, the algorithm is able to handle cases of intersectionality, i.e., of multi-dimensional discrimination of certain groups on grounds of several criteria. We discuss three main examples (credit applications; college admissions; insurance contracts) and map out the legal and policy implications of our approach. The explicit formalization of the trade-off between individual and group fairness allows this post-processing approach to be tailored to different situational contexts in which one or the other fairness criterion may take precedence. Finally, we evaluate our model experimentally.

A Mechanism for Allocating Benefits and Costs from Transmission Interconnections under Cooperation: A Case Study of the North Sea Offshore Grid

We propose a generic mechanism for allocating the benefits and costs that result from the development of international transmission interconnections under a cooperative agreement. The mechanism is based on a planning model that considers generation investments as a response to transmission developments, and the Shapley Value from cooperative game theory. This method provides a unique allocation of benefits and costs considering each country’s average incremental contribution to the cooperative agreement. The allocation satisfies an axiomatic definition of fairness. We demonstrate our results for three planned transmission interconnections in the North Sea and show that the proposed mechanism can be used as a basis for defining a set of Power Purchase Agreements among countries.

Efficient Performance-Centric Bandwidth Allocation with Fairness Tradeoff

Fair bandwidth allocation in datacenter networks has received a substantial amount of research attention, as multiple tenants are hosted by virtual machines in a public cloud. In the context of private datacenters, link bandwidth is shared among applications running data parallel frameworks, such as MapReduce, instead. In this paper, we introduce the rigorous definition of performance-centric fairness , with the guiding principle that the performance that data parallel applications will enjoy should be proportional to their weights. We first investigate the problem of maximizing application performance while maintaining strict performance-centric fairness. We then present an inherent tradeoff between fairness and efficiency, which is interpreted from the perspectives of bandwidth utilization and social welfare, respectively. From the first perspective, we propose an algorithm to improve bandwidth utilization by introducing an extended version of fairness. From the second perspective, we formulate an optimization problem of bandwidth allocation that maximizes the social welfare across all the applications, allowing a tunable degree of relaxation on performance-centric fairness. A distributed algorithm is then presented to solve the problem, based on dual based decomposition. With extensive simulations, we demonstrate the effectiveness of our algorithms in improving efficiency and application performance (by up to $1.4X$ ), with flexible degree of relaxation on the performance-centric fairness.

Integrating Technical and Legal Concepts of Privacy

We provide an overview of legal and technical concepts of privacy protection. Data protection guarantees exist in European Union at a primary law level since 2009, when the Lisbon Treaty has become effective. The new regulation called general data protection regulation adopted on April 27, 2016 will enter into effect on May 25, 2018. Regarding data protection, the US follows a very different approach. The cornerstone of the technical framework of data protection is the concept of differential privacy for which the privacy’s provable guarantees hold even against worst-case scenarios when the access to arbitrary auxiliary information is provided. Research directions for integrating legal and technical concepts of privacy are discussed. Although both fields, legal and technical, approach privacy from different perspectives, we expect that these approaches will converge to a common framework. The privacy-preserving framework, therefore, should be a balance between: 1) ambiguous legal concepts and precise mathematical notions of privacy; 2) the desired privacy level; 3) the information loss, which is measured by data utility metrics; and 4) the complexity and the practical feasibility of the proposed technique. Key notions of fairness are incompatible with each other, and hence any privacy-preserving framework should also include the trade-offs between conditions imposed in the definitions of fairness. The complex assembly of legal and technical components implementing various privacy requirements in the privacy-preserving framework, should be realized within the privacy-by-design concept.

The Neurobiology of Fairness and Social Justice: An Introduction

The study of social justice has always been an interdisciplinary undertaking, but in recent years neurobiologists have joined scholars and scientists from other areas to tackle complex questions concerning fairness, empathy, equality, hierarchy, and ideological conflict and polarization. By synthesizing insights from multiple, mutually informative levels of analysis, it is possible to shed new light on basic biological processes that reflect, inspire, and inhibit the pursuit of a more just society. With this special issue we highlight groundbreaking research on the neurobiology of fairness and social justice, bringing together six articles that address core themes of social justice, including individual variability in definitions of fairness, the genetic basis of economic egalitarianism, neural bases of empathy in environmental and intergroup domains, and the neural and genetic correlates of ideological polarization. Taken in conjunction, these diverse contributions bring multiple theoretical perspectives and research methods to bear on the shared goal of understanding and promoting social justice.

FairSquare: probabilistic verification of program fairness

With the range and sensitivity of algorithmic decisions expanding at a break-neck speed, it is imperative that we aggressively investigate fairness and bias in decision-making programs. First, we show that a number of recently proposed formal definitions of fairness can be encoded as probabilistic program properties. Second, with the goal of enabling rigorous reasoning about fairness, we design a novel technique for verifying probabilistic properties that admits a wide class of decision-making programs. Third, we present FairSquare, the first verification tool for automatically certifying that a program meets a given fairness property. We evaluate FairSquare on a range of decision-making programs. Our evaluation demonstrates FairSquare’s ability to verify fairness for a range of different programs, which we show are out-of-reach for state-of-the-art program analysis techniques.

Where Next for Long‐Term Care?

Where Next for Long‐Term Care?

Definition of QoE Fairness in Shared Systems

User-centric service and application management focuses on the quality of experience (QoE) as perceived by the end user. Thereby, the goal is to maximize QoE while ensuring fairness among users, e.g., for resource allocation and scheduling in shared systems. Although the literature suggests to consider consequently QoE fairness, there is currently no accepted definition of QoE fairness. The contribution of this letter is the definition of a generic QoE fairness index $F$ , which has desirable key properties as well as the rationale behind it. By using examples and a measurement study involving multiple users downloading web content over a bottleneck link, we differentiate the proposed index from QoS fairness and the widely used Jain’s fairness index. Based on results, we argue that neither QoS fairness nor Jain’s fairness index meet all of the desirable QoE-relevant Properties, which are met by $F$ . Consequently, the proposed index $F$ may be used to compare QoE fairness across systems and applications, thus serving as a benchmark for QoE management mechanisms and system optimization.