Fairness Objectives Research Articles

Reprint of: The multi-depot vehicle routing problem with profit fairness

One of the main obstacles in horizontal collaborative transportation is the proper distribution of gains among partners. In practice, a posteriori gain sharing mechanisms rarely guarantee that all partners feel treated fairly. In this paper, we present the multi-depot vehicle routing problem with profit fairness (MDVRP-PF), a bi-objective optimization problem that adds a fairness objective function to the classical cost minimization function. By studying the MDVRP-PF, we explore the effects of integrating fairness in the optimization process. To that end, we approximate the Pareto front of any problem instance using an adaptive large neighbourhood search algorithm, embedded within an ε-constraint scheme. This problem is solved for different instance types and time horizons. All the obtained Pareto fronts are characterized by six measures. Results show that the economic cost of fairness remains below 5%, even though remarkable differences exist among different instance types. We also observe that such costs tend to decrease when fairness is considered for longer time horizons.

Learning-Based Online QoE Optimization in Multi-Agent Video Streaming

Video streaming has become a major usage scenario for the Internet. The growing popularity of new applications, such as 4K and 360-degree videos, mandates that network resources must be carefully apportioned among different users in order to achieve the optimal Quality of Experience (QoE) and fairness objectives. This results in a challenging online optimization problem, as networks grow increasingly complex and the relevant QoE objectives are often nonlinear functions. Recently, data-driven approaches, deep Reinforcement Learning (RL) in particular, have been successfully applied to network optimization problems by modeling them as Markov decision processes. However, existing RL algorithms involving multiple agents fail to address nonlinear objective functions on different agents’ rewards. To this end, we leverage MAPG-finite, a policy gradient algorithm designed for multi-agent learning problems with nonlinear objectives. It allows us to optimize bandwidth distributions among multiple agents and to maximize QoE and fairness objectives on video streaming rewards. Implementing the proposed algorithm, we compare the MAPG-finite strategy with a number of baselines, including static, adaptive, and single-agent learning policies. The numerical results show that MAPG-finite significantly outperforms the baseline strategies with respect to different objective functions and in various settings, including both constant and adaptive bitrate videos. Specifically, our MAPG-finite algorithm maximizes QoE by 15.27% and maximizes fairness by 22.47% compared to the standard SARSA algorithm for a 2000 KB/s link.

LITMUS Predictor: An AI Assistant for Building Reliable, High-Performing and Fair Multilingual NLP Systems

Pre-trained multilingual language models are gaining popularity due to their cross-lingual zero-shot transfer ability, but these models do not perform equally well in all languages. Evaluating task-specific performance of a model in a large number of languages is often a challenge due to lack of labeled data, as is targeting improvements in low performing languages through few-shot learning. We present a tool - LITMUS Predictor - that can make reliable performance projections for a fine-tuned task-specific model in a set of languages without test and training data, and help strategize data labeling efforts to optimize performance and fairness objectives.

Integrating Behavioral, Economic, and Technical Insights to Understand and Address Algorithmic Bias: A Human-Centric Perspective

Many important decisions are increasingly being made with the help of information systems that use artificial intelligence and machine learning models. These computational models are designed to discover useful patterns from large amounts of data, which augment human capabilities to make decisions in various application domains. However, there are growing concerns regarding the ethics challenges faced by these automated decision-making (ADM) models, most notably on the issue of algorithmic bias , in which the models systematically produce less favorable (i.e., unfair) decisions for certain groups of people. In this commentary, we argue that algorithmic bias is not just a technical (e.g., computational or statistical) problem, and its successful resolution requires deep insights into individual and organizational behavior, economic incentives, as well as complex dynamics of the sociotechnical systems in which the ADM models are embedded. We discuss a human-centric, fairness-aware ADM framework that highlights the holistic involvement of human decision makers in each step of ADM. We review the emerging literature on fairness-aware machine learning and then discuss various strategic decisions that humans need to make, such as formulating proper fairness objectives, recognizing fairness-induced trade-offs and implications, utilizing machine learning model outputs, and managing/governing the decisions of ADM models. We further illustrate how these strategic decisions are jointly informed by behavioral, economic, and design sciences. Our discussions reveal a number of future research opportunities uniquely suitable for Management Information Systems (MIS) researchers to pursue.

Hybrid data analytic technique for grading fairness

PurposeFair grading produces learning ability levels that are understandable and acceptable to both learners and instructors. Norm-referenced grading can be achieved by several means such as z score, K-means and a heuristic. However, these methods typically deliver the varied degrees of grading fairness depending on input score data.Design/methodology/approachTo attain the fairest grading, this paper proposes a hybrid algorithm that integrates z score, K-means and heuristic methods with a novel fairness objective function as a decision function.FindingsDepending on an experimented data set, each of the algorithm's constituent methods could deliver the fairest grading results with fairness degrees ranging from 0.110 to 0.646. We also pointed out key factors in the fairness improvement of norm-referenced achievement grading.Originality/valueThe main contributions of this paper are four folds: the definition of fair norm-referenced grading requirements, a hybrid algorithm for fair norm-referenced grading, a fairness metric for norm-referenced grading and the fairness performance results of the statistical, heuristic and machine learning methods.

Proportionally Fair Organ Transplantation Allocation Rules

We introduce a new fairness measure for designing organ allocation rules for transplantation. In particular, we apply the proportional fairness measure for organ transplantation whose solution can be interpreted as the solution to the Nash bargaining problem for sharing limited donor organs among patients who seek to maximize their quality-adjusted life-years (QALYs). The motivation arises from several observations that current measures of fairness induce significant inefficiencies in terms of total QALYs of the patient population. We use the asymptotic results for the fluid approximation of a transplant queuing system to estimate the expected utility of patients and formulate an optimization problem where the decision maker (DM) partitions the set of organ types to achieve a proportionally fair objective. We use an achievable region approach to transform our formulation to an alternative optimization problem and show that the optimal allocation policy under the proportional fairness measure is assortative, which has the following insight: higher quality organs are allocated to patients who will have higher expected QALYs. We compare the performance of allocation rules developed for fairness purposes in organ allocation along with our proposed policy via a validated simulation model for heart transplantation in the US.

Ethics-by-design: efficient, fair and inclusive resource allocation using machine learning.

The distribution of crucial medical goods and services in conditions of scarcity is among the most important, albeit contested, areas of public policy development. Policymakers must strike a balance between multiple efficiency and fairness objectives, while reconciling disparate value judgments from a diverse set of stakeholders. We present a general framework for combining ethical theory, data modeling, and stakeholder input in this process and illustrate through a case study on designing organ transplant allocation policies. We develop a novel analytical tool, based on machine learning and optimization, designed to facilitate efficient and wide-ranging exploration of policy outcomes across multiple objectives. Such a tool enables all stakeholders, regardless of their technical expertise, to more effectively engage in the policymaking process by developing evidence-based value judgments based on relevant tradeoffs.

Harnessing UAVs for Fair 5G Bandwidth Allocation in Vehicular Communication via Deep Reinforcement Learning

Terrestrial infrastructure-based wireless networks do not always guarantee their resources will be shared uniformly by nodes in vehicular networks. This is due mainly to the uneven and dynamic geographical distribution of vehicles and path loss effects. In this paper, we leverage multiple fifth-generation (5G) unmanned aerial vehicles (UAVs) to enhance fairness in network resource allocation among vehicles by positioning UAVs on-demand as “flying communication infrastructure”. We propose a deep reinforcement learning (DRL) approach to determine UAVs’ position to improve network resource allocation fairness and efficiency while considering the UAVs’ flying range, communication range, and energy constraints. We use a parametric fairness function to attain a number of resource allocation objectives ranging from maximizing the total throughput of vehicles, maximizing minimum throughput, and achieving proportional bandwidth allocation. Simulation results show that the proposed DRL approach to UAV positioning can improve network resource allocation according to the targeted fairness objective.

Where to locate COVID-19 mass vaccination facilities?

The outbreak of COVID‐19 led to a record‐breaking race to develop a vaccine. However, the limited vaccine capacity creates another massive challenge: how to distribute vaccines to mitigate the near‐end impact of the pandemic? In the United States in particular, the new Biden administration is launching mass vaccination sites across the country, raising the obvious question of where to locate these clinics to maximize the effectiveness of the vaccination campaign. This paper tackles this question with a novel data‐driven approach to optimize COVID‐19 vaccine distribution. We first augment a state‐of‐the‐art epidemiological model, called DELPHI, to capture the effects of vaccinations and the variability in mortality rates across age groups. We then integrate this predictive model into a prescriptive model to optimize the location of vaccination sites and subsequent vaccine allocation. The model is formulated as a bilinear, nonconvex optimization model. To solve it, we propose a coordinate descent algorithm that iterates between optimizing vaccine distribution and simulating the dynamics of the pandemic. As compared to benchmarks based on demographic and epidemiological information, the proposed optimization approach increases the effectiveness of the vaccination campaign by an estimated 20%, saving an extra 4000 extra lives in the United States over a 3‐month period. The proposed solution achieves critical fairness objectives—by reducing the death toll of the pandemic in several states without hurting others—and is highly robust to uncertainties and forecast errors—by achieving similar benefits under a vast range of perturbations.

Mechanism Design for Fair and Efficient DSO Flexibility Markets

The proliferation of distributed energy assets necessitates the provision of flexibility to efficiently operate modern distribution systems. In this article, we propose a flexibility market through which the DSO may acquire flexibility services from asset aggregators in order to maintain network voltages and currents within safe limits. A max-min fair formulation is proposed for the allocation of flexibility. Since the DSO is not aware of each aggregator's local flexibility costs, we show that strategic misreporting can lead to severe loss of efficiency. Using mechanism design theory, we provide a mechanism that makes it a payoff-maximizing strategy for each aggregator to make truthful bids to the flexibility market. While typical truthful mechanisms only work when the objective is the maximization of Social Welfare, the proposed mechanism lets the DSO achieve incentive compatibility and optimality for the max-min fairness objective.

Managing Agricultural Water Considering Water Allocation Priority Based on Remote Sensing Data

To fairly distribute limited irrigation water resources in arid regions, a water allocation priority evaluation method based on remote sensing data was proposed and integrated with an optimization model. First, the water supply response unit was divided according to canal system conditions. Then, a spatialization method was used for generating spatial agricultural output value (income from planting industry) and grain yield (yield of food crops) with the help of NDVI and the potential yield of farmland. Third, the AHP-TOPSIS method was employed to calculate the water allocation priority based on the above information. Finally, the evaluation results were integrated with a nonlinear multiobjective model to optimally allocate agricultural land and water resources, considering the combined objective of minimum envy and proportional fairness. The method was applied to Hetao irrigation area, an arid agriculture-dominant region in Northwest China. After solving the model, optimization alternatives were obtained, which indicate that: (1) the spatial method of agricultural output value can improve the accuracy by around 16% compared with the traditional method, and the spatial method of grain yield also have good accuracy (MAPE = 14.66%); (2) the rank of water allocation priority can reflect more spatial information, and provide practical decision support for the distribution of water resources; (3) the envy index can better improve the efficiency of an allocation system compared to the Gini coefficient method.

A decision making framework for incorporating fairness in allocating slots at capacity-constrained airports

Fairness constitutes an important criterion for allocating scarce resources among self-interested parties. The capacity of congested airports is a scarce resource which is allocated in the form of slots to airlines to operate their flights. Traditionally, the optimisation of slot allocation efficiency has been considered as the primary criterion in making slot allocation decisions at capacity-constrained airports. The efficiency of slot allocation is expressed as a function of displacement, which is defined as the difference between the time requested by the airlines and the time allocated to airlines to operate their flights. Recent research work has considered fairness as an additional slot allocation objective. However, the literature has not adequately addressed the following two key research questions: i) in the presence of alternative fairness measures, which fairness measure should be used to investigate the slot allocation efficiency-fairness trade-off, and how the choice of the efficient frontier describing this trade-off should be made? and ii) given that the choice of the efficient frontier has been made, which slot allocation solution should be selected for implementation, and how this choice should be made? To address these questions, we investigate three inter-airline fairness objectives, and we introduce a decision-making framework to assist decision-makers in identifying the most preferable acceptable slot allocation outcome. The proposed decision-making framework is based on a voting mechanism which incorporates the preferences of stakeholders. We use slot request and capacity data to determine the most preferable slot allocation outcome at a capacity-constrained airport.

Causal Sets in Antidiscrimination Law

Standards of causation in antidiscrimination law, and disparate-treatment cases in particular, are deeply flawed. Their defects have caused an illogical, obscure, and unworkable proof scheme that requires an overhaul to curb the harm that it engenders and to allow the antidiscrimination statutes to serve their objectives effectively. This Article proposes a theory and method of causation that achieves this goal. The problem stems from the inadequacies associated with current standards of causation in disparate-treatment cases—the but-for test and the motivating-factor test. The proposed “factorial” approach introduces a causal standard that addresses these inadequacies. It entails three innovations over current causation schemes: 1) it adopts a predominant framework for cause and effect in the sciences, called the “potential-outcomes” framework, as a central structure in which to sharply define and analyze the causal inquiry; 2) it employs a causal measure, called the “NESS” test, that refines and, in a sense, unifies the but-for and motivating-factor tests by retaining the central feature of the but-for test—the “necessity condition”—but in a less restrictive form; and 3) it applies a legal framework grounded in tort law and recent advances regarding multiple sufficient causes. In addition to reflecting actual cause and effect, the proposed approach promotes antidiscrimination law’s deterrence and fairness objectives, and it allows an interpretation of causal language in antidiscrimination statutes that is consistent with good policy and Congress’s intent—an interpretation not possible under current standards.

Multi-Agent Deep Reinforcement Learning Multiple Access for Heterogeneous Wireless Networks With Imperfect Channels

This paper investigates a futuristic spectrum sharing paradigm for heterogeneous wireless networks with imperfect channels. In the heterogeneous networks, multiple wireless networks adopt different medium access control (MAC) protocols to share a common wireless spectrum and each network is unaware of the MACs of others. This paper aims to design a distributed deep reinforcement learning (DRL) based MAC protocol for a particular network, and the objective of this network is to achieve a global <inline-formula><tex-math notation="LaTeX">$\alpha$</tex-math></inline-formula> -fairness objective. In the conventional DRL framework, feedback/reward given to the agent is always correctly received, so that the agent can optimize its strategy based on the received reward. In our wireless application where the channels are noisy, the feedback/reward (i.e., the ACK packet) may be lost due to channel noise and interference. Without correct feedback, the agent (i.e., the network user) may fail to find a good solution. Moreover, in the distributed protocol, each agent makes decisions on its own. It is a challenge to guarantee that the multiple agents will make coherent decisions and work together to achieve the same objective, particularly in the face of imperfect feedback channels. To tackle the challenge, we put forth (i) a feedback recovery mechanism to recover missing feedback information, and (ii) a two-stage action selection mechanism to aid coherent decision making to reduce transmission collisions among the agents. Extensive simulation results demonstrate the effectiveness of these two mechanisms. Last but not least, we believe that the feedback recovery mechanism and the two-stage action selection mechanism can also be used in general distributed multi-agent reinforcement learning problems in which feedback information on rewards can be corrupted.

Personnel Scheduling Problem under Hierarchical Management Based on Intelligent Algorithm

This paper studies a special scheduling problem under hierarchical management in nurse staff. This is a more complex rostering problem than traditional nurse scheduling. The first is that the rostering requirements of charge nurses and general nurses are different under hierarchical management. The second is that nurses are preferable for relative fair rather than absolute fair under hierarchical management. The model aims at allocating the required workload to meet the operational requirements, weekend rostering preferences, and relative fairness preferences. Two hybrid heuristic algorithms based on multiobjective grey wolf optimizer (MOGWO) and three corresponding single heuristic algorithms are employed to solve this problem. The experimental results based on real cases from the Third People’s Hospital, Panzhihua, China, show that MOGWO does not as good as it does on other engineering optimization. However, the hybrid algorithms based on MOGWO are better than corresponding single algorithms on generational distance (GD) and spacing (SP) of Pareto solutions. Furthermore, for relative fair rostering objective, NSGAII-MOGWO has more power to find the optimal solution in the dimension of relative fairness.

Subject of authority as a plaintiff in administrative cases

This scientific article examines the subject of power in the administrative process on the basis of Ukrainian legislation and practice of the European Court of Human Rights. In accordance with the tasks of administrative proceedings, the subject of power is mainly in the procedural status of the defendant. In administrative proceedings, public and private interests are opposed, the powers of justice are combined and the authority of the public authority is responsible for the control of the law of the public authority. In a dispute between a citizen and the state, a fair objective assessment of the actions of a public authority is of particular importance for a judge, taking into account the priority of human and civil rights and freedoms. To date, a number of issues remain relevant that determine the grounds for applying to the court of a subject of authority, its status and the level of parity between the plaintiff and the defendant in the administrative process. The article is aimed at establishing the legal status of the subject of power in the system of administrative proceedings through the prism of protection and the performance of delegated functions. The subject of power is classified on the grounds of belonging to the authorities and management and in accordance with the official functions aimed at the exercise of power. Acts in court mainly as a defendant party in an administrative process, proving the legality of its actions or inaction in the case. According to the law, only in some cases acts as a plaintiff in administrative cases. The task of administrative proceedings in such cases is achieved by eliminating the threat of unjustified interference and violations the of the rights of the subject of power. The subject of power has the right to appeal to the administrative court only in cases specified in the Constitution and the laws of Ukraine and in order to implement the powers vested in it. Judicial practice indicates that the authorities are endowed with authority not acting on an equal footing with other participants in the relevant relations and perform managerial functions aimed at providing state guarantees. In court cases where the subject of power acts as a plaintiff to the subject of private law, the court is guided by the principle of proportionality as a universal general legal system aimed at ensuring a balance of private and public interests in the legal regulation.

Fair Value Opinion Shopping or Objective Valuation?

The literature suggests that fair value (FV) estimates from external third-party sources are more reliable than those from managerial inputs. However, using a sample of private insurance companies, we provide evidence that even verifiable FV estimates derived from external third-party sources are not immune to managerial opportunism due to the discretion insurers have in switching sources. We posit that such source switches could either be driven by managerial incentive to faithfully report FV (objective valuation) or to inflate FV estimates to avoid other-than-temporary impairment (OTTI) (FV opinion shopping). Our results indicate FV opinion shopping is the more dominant motive and that source-switches yield less accurate and more upwardly biased FV estimates. Upward switches also reduce both the likelihood and magnitude of OTTI especially for high-impairment-risk securities. Overall, our evidence suggests that opportunism with respect to source-switching potentially compromises the disciplining role of independent third parties in the valuation of fair-valued assets.

Community solar photovoltaic service strategy for commercial buildings considering profit balancing and fairness

A community solar photovoltaic (PV) system is an effective way to save energy bills in commercial buildings. In such systems, the main problems are how to serve the shared resources considering profit balancing and profit-sharing fairness. In this study, a community solar PV service strategy for commercial buildings is introduced, which does not only pursue profit maximization for the entire community, but also considers two additional issues, namely, the profit balancing between the PV service provider (SP) and participants and the profit-sharing fairness among the participants. A profit-balancing factor is defined as a representation of the nature of the SP, and the profit maximization model, considering fair profit-sharing, is applied. By applying the Lagrangian relaxation and dual ascent method, the optimal solution is achieved in an iterative manner. A case study shows that participants can save 50 and 40% of their total energy bill when the service is operated with the objective of profit maximization and proportional fairness, respectively. The proportional-fairness operation enhances the Jain’s fairness index up to 51% compared with the profit-maximization operation. The effects on the profit and fairness for various service methods are explored to provide guidance in designing community solar PV services.

Ambassador of Lifelong Learning/ Public Libraries in Multicultural Society: The Case of Pakistan

Cultural diversity is an asset, an added value, and an opportunity. Public libraries promote cultural diversity and intercultural dialogue. They play dynamic roles as social meeting places. Public libraries in a culturally diverse society work as mediators between society, culture and users. Pakistani community is a mixture of different languages and cultures which have different dimensions’ and local cultural roots. This present research aims to draw a real picture of public library's role in the multicult ural and digital society of Pakistani. The role of public libraries can be very vital. They can play a critical part in advocating multiplicity responsiveness, as they are responsible to provide resources to the public. They provide help and guide general public to accomplish the objective of social and monetary fairness and equity that at last will prompt a peaceful & flourishing society. To get the maximum benefits from the cultural awareness programs devoted efforts, sufficient funds, adequate training a nd total commitment are required from the top management to low management of public libraries of Lahore. Public libraries can play essential role to overcome the cultural gap in Pakistani society. The work serves as a source to develop concepts and best practices of multicultural library work for all librarians concerned with these issues. It will provide an opportunity for professionals, students, and library members to create new awareness and practice.

Modeling and Simulation of Collaborative Dispatching of Disaster Relief Materials Based on Urgency

In order to address the lack of collaborative decision and failure to notice the emergency and fairness of relief after disasters have occurred, a collaborative decision-making system for emergency relief materials dispatching is established. According to the forecast of the demand for postdisaster relief materials, the entropy weight-TOPSIS method is applied to measure the urgency of the disaster area; then, a “Hub-and-Spoke” dispatching network is constructed. In this paper, a multiobjective collaborative relief material dispatching model is built, which has great performance in terms of minimal distribution cost and maximal fairness, and the objective of fairness requires minimizing the penalty cost caused by unsatisfied demands. Based on the urgency of demand points, the simulated annealing algorithm is designed to solve the Pareto disaggregation of multiobjective optimization model. The performance of the model is verified with the case of Wenchuan Earthquake. The results indicate that if the fair distribution of supplies is emphasized, it will increase the number of rescue vehicles and the number of distribution batches. On the other hand, a variety of relief material dispatching plans can be provided based on calculation of the Pareto front for policy-makers.