Group fairness metrics are an established way of assessing the fairness of prediction-based decision-making systems. However, these metrics are still insufficiently linked to philosophical theories, and their moral meaning is often unclear. In this paper, we propose a comprehensive framework for group fairness metrics, which links them to more theories of distributive justice. The different group fairness metrics differ in their choices about how to measure the benefit or harm of a decision for the affected individuals, and what moral claims to benefits are assumed. Our unifying framework reveals the normative choices associated with standard group fairness metrics and allows an interpretation of their moral substance. In addition, this broader view provides a structure for the expansion of standard fairness metrics that we find in the literature. This expansion allows addressing several criticisms of standard group fairness metrics, specifically: (1) they are parity-based, i.e., they demand some form of equality between groups, which may sometimes be detrimental to marginalized groups; (2) they only compare decisions across groups but not the resulting consequences for these groups; and (3) the full breadth of the distributive justice literature is not sufficiently represented.
相關內容
With the introduction of machine learning in high-stakes decision making, ensuring algorithmic fairness has become an increasingly important problem to solve. In response to this, many mathematical definitions of fairness have been proposed, and a variety of optimisation techniques have been developed, all designed to maximise a defined notion of fairness. However, fair solutions are reliant on the quality of the training data, and can be highly sensitive to noise. Recent studies have shown that robustness (the ability for a model to perform well on unseen data) plays a significant role in the type of strategy that should be used when approaching a new problem and, hence, measuring the robustness of these strategies has become a fundamental problem. In this work, we therefore propose a new criterion to measure the robustness of various fairness optimisation strategies - the robustness ratio. We conduct multiple extensive experiments on five bench mark fairness data sets using three of the most popular fairness strategies with respect to four of the most popular definitions of fairness. Our experiments empirically show that fairness methods that rely on threshold optimisation are very sensitive to noise in all the evaluated data sets, despite mostly outperforming other methods. This is in contrast to the other two methods, which are less fair for low noise scenarios but fairer for high noise ones. To the best of our knowledge, we are the first to quantitatively evaluate the robustness of fairness optimisation strategies. This can potentially can serve as a guideline in choosing the most suitable fairness strategy for various data sets.
A variety of different performance metrics are commonly used in the machine learning literature for the evaluation of classification systems. Some of the most common ones for measuring quality of hard decisions are standard and balanced accuracy, standard and balanced error rate, F-beta score, and Matthews correlation coefficient (MCC). In this document, we review the definition of these and other metrics and compare them with the expected cost (EC), a metric introduced in every statistical learning course but rarely used in the machine learning literature. We show that both the standard and balanced error rates are special cases of the EC. Further, we show its relation with F-score and MCC and argue that EC is superior to these traditional metrics, being more elegant, general, and intuitive, as well as being based on basic principles from statistics. The metrics above measure the quality of hard decisions. Yet, most modern classification systems output continuous scores for the classes which we may want to evaluate directly. Metrics for measuring the quality of system scores include the area under the ROC curve, equal error rate, cross-entropy, Brier score, and Bayes EC or Bayes risk, among others. The last three metrics are special cases of a family of metrics given by the expected value of proper scoring rules (PSRs). We review the theory behind these metrics and argue that they are the most principled way to measure the quality of the posterior probabilities produced by a system. Finally, we show how to use these metrics to compute the system's calibration loss and compare this metric with the standard expected calibration error (ECE), arguing that calibration loss based on PSRs is superior to the ECE for a variety of reasons.
Current WHO guidelines set prevalence thresholds below which a Neglected Tropical Disease can be considered to have been eliminated as a public health problem, and specify how surveys to assess whether elimination has been achieved should be designed and analysed, based on classical survey sampling methods. In this paper we describe an alternative approach based on geospatial statistical modelling. We first show the gains in efficiency that can be obtained by exploiting any spatial correlation in the underlying prevalence surface. We then suggest that the current guidelines implicit use of a significance testing argument is not appropriate; instead, we argue for a predictive inferential framework, leading to design criteria based on controlling the rates at which areas whose true prevalence lies above and below the elimination threshold are incorrectly classified. We describe how this approach naturally accommodates context-specific information in the form of georeferenced covariates that have been shown to be predictive of disease prevalence. Finally, we give a progress report of an ongoing collaboration with the Guyana Ministry of Health Neglected Tropical Disease program on the design of an IDA (Ivermectin, Diethylcarbamazine and Albendazole) Impact Survey (IIS) of lymphatic filariasis to be conducted in Guyana in early 2023
Predicting the number of defects in a project is critical for project test managers to allocate budget, resources, and schedule for testing, support and maintenance efforts. Software Defect Prediction models predict the number of defects in given projects after training the model with historical defect related information. The majority of defect prediction studies focused on predicting defect-prone modules from methods, and class-level static information, whereas this study predicts defects from project-level information based on a cross-company project dataset. This study utilizes software sizing metrics, effort metrics, and defect density information, and focuses on developing defect prediction models that apply various machine learning algorithms. One notable issue in existing defect prediction studies is the lack of transparency in the developed models. Consequently, the explain-ability of the developed model has been demonstrated using the state-of-the-art post-hoc model-agnostic method called Shapley Additive exPlanations (SHAP). Finally, important features for predicting defects from cross-company project information were identified.
The online advertising market has recently reached the 500 billion dollar mark, and to accommodate the need to match a user with the highest bidder at a fraction of a second, it has moved towards a complex automated model involving numerous agents and middle men. Stimulated by potential revenue and the lack of transparency, bad actors have found ways to abuse it, circumvent restrictions, and generate substantial revenue from objectionable and even illegal content. To make matters worse, they often receive advertisements from respectable companies which have nothing to do with these illegal activities. Altogether, advertiser money is funneled towards unknown entities, supporting their objectionable operations and maintaining their existence. In this project, we work towards understanding the extent of the problem and shed light on how shady agents take advantage of gaps in the ad ecosystem to monetize their operations. We study over 7 million websites and examine how state-of-the-art standards associated with online advertising are applied. We discover and present actual practices observed in the wild and show that publishers are able to monetize objectionable and illegal content and generate thousands of dollars of revenue on a monthly basis.
Empirical evidence suggests that algorithmic decisions driven by Machine Learning (ML) techniques threaten to discriminate against legally protected groups or create new sources of unfairness. This work supports the contextual approach to fairness in EU non-discrimination legal framework and aims at assessing up to what point we can assure legal fairness through fairness metrics and under fairness constraints. For that, we analyze the legal notion of non-discrimination and differential treatment with the fairness definition Demographic Parity (DP) through Conditional Demographic Disparity (CDD). We train and compare different classifiers with fairness constraints to assess whether it is possible to reduce bias in the prediction while enabling the contextual approach to judicial interpretation practiced under EU non-discrimination laws. Our experimental results on three scenarios show that the in-processing bias mitigation algorithm leads to different performances in each of them. Our experiments and analysis suggest that AI-assisted decision-making can be fair from a legal perspective depending on the case at hand and the legal justification. These preliminary results encourage future work which will involve further case studies, metrics, and fairness notions.
We consider a cross-silo federated learning (FL) setting where a machine learning model with a fully connected first layer is trained between different clients and a central server using FedAvg, and where the aggregation step can be performed with secure aggregation (SA). We present SRATTA an attack relying only on aggregated models which, under realistic assumptions, (i) recovers data samples from the different clients, and (ii) groups data samples coming from the same client together. While sample recovery has already been explored in an FL setting, the ability to group samples per client, despite the use of SA, is novel. This poses a significant unforeseen security threat to FL and effectively breaks SA. We show that SRATTA is both theoretically grounded and can be used in practice on realistic models and datasets. We also propose counter-measures, and claim that clients should play an active role to guarantee their privacy during training.
It is often of interest to assess whether a function-valued statistical parameter, such as a density function or a mean regression function, is equal to any function in a class of candidate null parameters. This can be framed as a statistical inference problem where the target estimand is a scalar measure of dissimilarity between the true function-valued parameter and the closest function among all candidate null values. These estimands are typically defined to be zero when the null holds and positive otherwise. While there is well-established theory and methodology for performing efficient inference when one assumes a parametric model for the function-valued parameter, methods for inference in the nonparametric setting are limited. When the null holds, and the target estimand resides at the boundary of the parameter space, existing nonparametric estimators either achieve a non-standard limiting distribution or a sub-optimal convergence rate, making inference challenging. In this work, we propose a strategy for constructing nonparametric estimators with improved asymptotic performance. Notably, our estimators converge at the parametric rate at the boundary of the parameter space and also achieve a tractable null limiting distribution. As illustrations, we discuss how this framework can be applied to perform inference in nonparametric regression problems, and also to perform nonparametric assessment of stochastic dependence.
In this paper we will provide a quantitative analysis of a simple model of the Federal Disaster Assistance policy from the viewpoint of three different stakeholders. This quantitative methodology is new and has applications to other areas such as business and healthcare processes. The stakeholders are interested in process transparency but each has a different opinion on precisely what constitutes transparency. We will also consider three modifications to the Federal Disaster Assistance policy and analyse, from a stakeholder viewpoint, how stakeholder satisfaction changes from process to process. This analysis is used to rank the favourability of four policies with respect to all collective stakeholder preferences.
This paper focuses on the expected difference in borrower's repayment when there is a change in the lender's credit decisions. Classical estimators overlook the confounding effects and hence the estimation error can be magnificent. As such, we propose another approach to construct the estimators such that the error can be greatly reduced. The proposed estimators are shown to be unbiased, consistent, and robust through a combination of theoretical analysis and numerical testing. Moreover, we compare the power of estimating the causal quantities between the classical estimators and the proposed estimators. The comparison is tested across a wide range of models, including linear regression models, tree-based models, and neural network-based models, under different simulated datasets that exhibit different levels of causality, different degrees of nonlinearity, and different distributional properties. Most importantly, we apply our approaches to a large observational dataset provided by a global technology firm that operates in both the e-commerce and the lending business. We find that the relative reduction of estimation error is strikingly substantial if the causal effects are accounted for correctly.
小貼士
登錄享
相關主題
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191