arXiv subjects

Aleksandra Korolova

Publications and source records attributed to Aleksandra Korolova.

At least 37 records · Page 2Linked to original sources

arXiv2024

Auditing for Racial Discrimination in the Delivery of Education Ads

Digital ads on social-media platforms play an important role in shaping access to economic opportunities. Our work proposes and implements a new third-party auditing method that can evaluate racial bias in the delivery of ads for education opportunities. Third-party auditing is important because it allows external parties to demonstrate presence or absence of bias in social-media algorithms. Education is a domain with legal protections against discrimination and concerns of racial-targeting, but bias induced by ad delivery algorithms has not been previously explored in this domain. Prior audits demonstrated discrimination in platforms' delivery of ads to users for housing and employment ads. These audit findings supported legal action that prompted Meta to change their ad-delivery algorithms to reduce bias, but only in the domains of housing, employment, and credit. In this work, we propose a new methodology that allows us to measure racial discrimination in a platform's ad delivery algorithms for education ads. We apply our method to Meta using ads for real schools and observe the results of delivery. We find evidence of racial discrimination in Meta's algorithmic delivery of ads for education opportunities, posing legal and ethical concerns. Our results extend evidence of algorithmic discrimination to the education domain, showing that current bias mitigation mechanisms are narrow in scope, and suggesting a broader role for third-party auditing of social media in areas where ensuring non-discrimination is important.

cs.CY↗

arXiv2024

Stability and Multigroup Fairness in Ranking with Uncertain Predictions

Rankings are ubiquitous across many applications, from search engines to hiring committees. In practice, many rankings are derived from the output of predictors. However, when predictors trained for classification tasks have intrinsic uncertainty, it is not obvious how this uncertainty should be represented in the derived rankings. Our work considers ranking functions: maps from individual predictions for a classification task to distributions over rankings. We focus on two aspects of ranking functions: stability to perturbations in predictions and fairness towards both individuals and subgroups. Not only is stability an important requirement for its own sake, but -- as we show -- it composes harmoniously with individual fairness in the sense of Dwork et al. (2012). While deterministic ranking functions cannot be stable aside from trivial scenarios, we show that the recently proposed uncertainty aware (UA) ranking functions of Singh et al. (2021) are stable. Our main result is that UA rankings also achieve multigroup fairness through successful composition with multiaccurate or multicalibrated predictors. Our work demonstrates that UA rankings naturally interpolate between group and individual level fairness guarantees, while simultaneously satisfying stability guarantees important whenever machine-learned predictions are used.

cs.LG↗

arXiv2023

Fairness in Matching under Uncertainty

The prevalence and importance of algorithmic two-sided marketplaces has drawn attention to the issue of fairness in such settings. Algorithmic decisions are used in assigning students to schools, users to advertisers, and applicants to job interviews. These decisions should heed the preferences of individuals, and simultaneously be fair with respect to their merits (synonymous with fit, future performance, or need). Merits conditioned on observable features are always \emph{uncertain}, a fact that is exacerbated by the widespread use of machine learning algorithms to infer merit from the observables. As our key contribution, we carefully axiomatize a notion of individual fairness in the two-sided marketplace setting which respects the uncertainty in the merits; indeed, it simultaneously recognizes uncertainty as the primary potential cause of unfairness and an approach to address it. We design a linear programming framework to find fair utility-maximizing distributions over allocations, and we show that the linear program is robust to perturbations in the estimated parameters of the uncertain merit distributions, a key property in combining the approach with machine learning techniques.

cs.LG↗

arXiv2023

Discrimination through Image Selection by Job Advertisers on Facebook

Targeted advertising platforms are widely used by job advertisers to reach potential employees; thus issues of discrimination due to targeting that have surfaced have received widespread attention. Advertisers could misuse targeting tools to exclude people based on gender, race, location and other protected attributes from seeing their job ads. In response to legal actions, Facebook disabled the ability for explicit targeting based on many attributes for some ad categories, including employment. Although this is a step in the right direction, prior work has shown that discrimination can take place not just due to the explicit targeting tools of the platforms, but also due to the impact of the biased ad delivery algorithm. Thus, one must look at the potential for discrimination more broadly, and not merely through the lens of the explicit targeting tools. In this work, we propose and investigate the prevalence of a new means for discrimination in job advertising, that combines both targeting and delivery -- through the disproportionate representation or exclusion of people of certain demographics in job ad images. We use the Facebook Ad Library to demonstrate the prevalence of this practice through: (1) evidence of advertisers running many campaigns using ad images of people of only one perceived gender, (2) systematic analysis for gender representation in all current ad campaigns for truck drivers and nurses, (3) longitudinal analysis of ad campaign image use by gender and race for select advertisers. After establishing that the discrimination resulting from a selective choice of people in job ad images, combined with algorithmic amplification of skews by the ad delivery algorithm, is of immediate concern, we discuss approaches and challenges for addressing it.

cs.CY↗

arXiv2023

Having your Privacy Cake and Eating it Too: Platform-supported Auditing of Social Media Algorithms for Public Interest

Social media platforms curate access to information and opportunities, and so play a critical role in shaping public discourse today. The opaque nature of the algorithms these platforms use to curate content raises societal questions. Prior studies have used black-box methods to show that these algorithms can lead to biased or discriminatory outcomes. However, existing auditing methods face fundamental limitations because they function independent of the platforms. Concerns of potential harm have prompted proposal of legislation in both the U.S. and the E.U. to mandate a new form of auditing where vetted external researchers get privileged access to social media platforms. Unfortunately, to date there have been no concrete technical proposals to provide such auditing, because auditing at scale risks disclosure of users' private data and platforms' proprietary algorithms. We propose a new method for platform-supported auditing that can meet the goals of the proposed legislation. Our first contribution is to enumerate the challenges of existing auditing methods to implement these policies at scale. Second, we suggest that limited, privileged access to relevance estimators is the key to enabling generalizable platform-supported auditing by external researchers. Third, we show platform-supported auditing need not risk user privacy nor disclosure of platforms' business interests by proposing an auditing framework that protects against these risks. For a particular fairness metric, we show that ensuring privacy imposes only a small constant factor increase (6.34x as an upper bound, and 4x for typical parameters) in the number of samples required for accurate auditing. Our technical contributions, combined with ongoing legal and policy efforts, can enable public oversight into how social media platforms affect individuals and society by moving past the privacy-vs-transparency hurdle.

cs.CY↗

arXiv2023

Pushing the Boundaries of Private, Large-Scale Query Answering

We address the problem of efficiently and effectively answering large numbers of queries on a sensitive dataset while ensuring differential privacy (DP). We separately analyze this problem in two distinct settings, grounding our work in a state-of-the-art DP mechanism for large-scale query answering: the Relaxed Adaptive Projection (RAP) mechanism. The first setting is a classic setting in DP literature where all queries are known to the mechanism in advance. Within this setting, we identify challenges in the RAP mechanism's original analysis, then overcome them with an enhanced implementation and analysis. We then extend the capabilities of the RAP mechanism to be able to answer a more general and powerful class of queries (r-of-k thresholds) than previously considered. Empirically evaluating this class, we find that the mechanism is able to answer orders of magnitude larger sets of queries than prior works, and does so quickly and with high utility. We then define a second setting motivated by real-world considerations and whose definition is inspired by work in the field of machine learning. In this new setting, a mechanism is only given partial knowledge of queries that will be posed in the future, and it is expected to answer these future-posed queries with high utility. We formally define this setting and how to measure a mechanism's utility within it. We then comprehensively empirically evaluate the RAP mechanism's utility within this new setting. From this evaluation, we find that even with weak partial knowledge of the future queries that will be posed, the mechanism is able to efficiently and effectively answer arbitrary queries posed in the future. Taken together, the results from these two settings advance the state of the art on differentially private large-scale query answering.

cs.CR↗

arXiv2023

"You Can't Fix What You Can't Measure": Privately Measuring Demographic Performance Disparities in Federated Learning

As in traditional machine learning models, models trained with federated learning may exhibit disparate performance across demographic groups. Model holders must identify these disparities to mitigate undue harm to the groups. However, measuring a model's performance in a group requires access to information about group membership which, for privacy reasons, often has limited availability. We propose novel locally differentially private mechanisms to measure differences in performance across groups while protecting the privacy of group membership. To analyze the effectiveness of the mechanisms, we bound their error in estimating a disparity when optimized for a given privacy budget. Our results show that the error rapidly decreases for realistic numbers of participating clients, demonstrating that, contrary to what prior work suggested, protecting privacy is not necessarily in conflict with identifying performance disparities of federated models.

cs.LG↗

arXiv2021

Robust Allocations with Diversity Constraints

We consider the problem of allocating divisible items among multiple agents, and consider the setting where any agent is allowed to introduce diversity constraints on the items they are allocated. We motivate this via settings where the items themselves correspond to user ad slots or task workers with attributes such as race and gender on which the principal seeks to achieve demographic parity. We consider the following question: When an agent expresses diversity constraints into an allocation rule, is the allocation of other agents hurt significantly? If this happens, the cost of introducing such constraints is disproportionately borne by agents who do not benefit from diversity. We codify this via two desiderata capturing robustness. These are no negative externality -- other agents are not hurt -- and monotonicity -- the agent enforcing the constraint does not see a large increase in value. We show in a formal sense that the Nash Welfare rule that maximizes product of agent values is uniquely positioned to be robust when diversity constraints are introduced, while almost all other natural allocation rules fail this criterion. We also show that the guarantees achieved by Nash Welfare are nearly optimal within a widely studied class of allocation rules. We finally perform an empirical simulation on real-world data that models ad allocations to show that this gap between Nash Welfare and other rules persists in the wild.

cs.GT↗

arXiv2021

Auditing for Discrimination in Algorithms Delivering Job Ads

Ad platforms such as Facebook, Google and LinkedIn promise value for advertisers through their targeted advertising. However, multiple studies have shown that ad delivery on such platforms can be skewed by gender or race due to hidden algorithmic optimization by the platforms, even when not requested by the advertisers. Building on prior work measuring skew in ad delivery, we develop a new methodology for black-box auditing of algorithms for discrimination in the delivery of job advertisements. Our first contribution is to identify the distinction between skew in ad delivery due to protected categories such as gender or race, from skew due to differences in qualification among people in the targeted audience. This distinction is important in U.S. law, where ads may be targeted based on qualifications, but not on protected categories. Second, we develop an auditing methodology that distinguishes between skew explainable by differences in qualifications from other factors, such as the ad platform's optimization for engagement or training its algorithms on biased data. Our method controls for job qualification by comparing ad delivery of two concurrent ads for similar jobs, but for a pair of companies with different de facto gender distributions of employees. We describe the careful statistical tests that establish evidence of non-qualification skew in the results. Third, we apply our proposed methodology to two prominent targeted advertising platforms for job ads: Facebook and LinkedIn. We confirm skew by gender in ad delivery on Facebook, and show that it cannot be justified by differences in qualifications. We fail to find skew in ad delivery on LinkedIn. Finally, we suggest improvements to ad platform practices that could make external auditing of their algorithms in the public interest more feasible and accurate.

cs.CY↗

arXiv2021

Advances and Open Problems in Federated Learning

Federated learning (FL) is a machine learning setting where many clients (e.g. mobile devices or whole organizations) collaboratively train a model under the orchestration of a central server (e.g. service provider), while keeping the training data decentralized. FL embodies the principles of focused data collection and minimization, and can mitigate many of the systemic privacy risks and costs resulting from traditional, centralized machine learning and data science approaches. Motivated by the explosive growth in FL research, this paper discusses recent advances and presents an extensive collection of open problems and challenges.

cs.LG↗

arXiv2020

The Power of The Hybrid Model for Mean Estimation

We explore the power of the hybrid model of differential privacy (DP), in which some users desire the guarantees of the local model of DP and others are content with receiving the trusted-curator model guarantees. In particular, we study the utility of hybrid model estimators that compute the mean of arbitrary real-valued distributions with bounded support. When the curator knows the distribution's variance, we design a hybrid estimator that, for realistic datasets and parameter settings, achieves a constant factor improvement over natural baselines. We then analytically characterize how the estimator's utility is parameterized by the problem setting and parameter choices. When the distribution's variance is unknown, we design a heuristic hybrid estimator and analyze how it compares to the baselines. We find that it often performs better than the baselines, and sometimes almost as well as the known-variance estimator. We then answer the question of how our estimator's utility is affected when users' data are not drawn from the same distribution, but rather from distributions dependent on their trust model preference. Concretely, we examine the implications of the two groups' distributions diverging and show that in some cases, our estimators maintain fairly high utility. We then demonstrate how our hybrid estimator can be incorporated as a sub-component in more complex, higher-dimensional applications. Finally, we propose a new privacy amplification notion for the hybrid model that emerges due to interaction between the groups, and derive corresponding amplification results for our hybrid estimators.

cs.CR↗

arXiv2020

On the Data Fight Between Cities and Mobility Providers

E-Scooters are changing transportation habits. In an attempt to oversee scooter usage, the Los Angeles Department of Transportation has put forth a specification that requests detailed data on scooter usage from scooter companies. In this work, we first argue that L.A.'s data request for using a new specification is not warranted as proposed use cases can be met by already existing specifications. Second, we show that even the existing specification, that requires companies to publish real-time data of parked scooters, puts the privacy of individuals using the scooters at risk. We then propose an algorithm that enables formal privacy and utility guarantees when publishing parked scooters data, allowing city authorities to meet their use cases while preserving riders' privacy.

cs.CY↗

arXiv2019

The power of synergy in differential privacy: Combining a small curator with local randomizers

Motivated by the desire to bridge the utility gap between local and trusted curator models of differential privacy for practical applications, we initiate the theoretical study of a hybrid model introduced by "Blender" [Avent et al.,\ USENIX Security '17], in which differentially private protocols of n agents that work in the local-model are assisted by a differentially private curator that has access to the data of m additional users. We focus on the regime where m << n and study the new capabilities of this (m,n)-hybrid model. We show that, despite the fact that the hybrid model adds no significant new capabilities for the basic task of simple hypothesis-testing, there are many other tasks (under a wide range of parameters) that can be solved in the hybrid model yet cannot be solved either by the curator or by the local-users separately. Moreover, we exhibit additional tasks where at least one round of interaction between the curator and the local-users is necessary -- namely, no hybrid model protocol without such interaction can solve these tasks. Taken together, our results show that the combination of the local model with a small curator can become part of a promising toolkit for designing and implementing differential privacy.

cs.DS↗

arXiv2019

Ad Delivery Algorithms: The Hidden Arbiters of Political Messaging

Political campaigns are increasingly turning to digital advertising to reach voters. These platforms empower advertisers to target messages to platform users with great precision, including through inferences about those users' political affiliations. However, prior work has shown that platforms' ad delivery algorithms can selectively deliver ads within these target audiences in ways that can lead to demographic skews along race and gender lines, often without an advertiser's knowledge. In this study, we investigate the impact of Facebook's ad delivery algorithms on political ads. We run a series of political ads on Facebook and measure how Facebook delivers those ads to different groups, depending on an ad's content (e.g., the political viewpoint featured) and targeting criteria. We find that Facebook's ad delivery algorithms effectively differentiate the price of reaching a user based on their inferred political alignment with the advertised content, inhibiting political campaigns' ability to reach voters with diverse political views. This effect is most acute when advertisers use small budgets, as Facebook's delivery algorithm tends to preferentially deliver to the users who are, according to Facebook's estimation, most relevant. Our findings point to advertising platforms' potential role in political polarization and creating informational filter bubbles. Furthermore, some large ad platforms have recently changed their policies to restrict the targeting tools they offer to political campaigns; our findings show that such reforms will be insufficient if the goal is to ensure that political ads are shown to users of diverse political views. Our findings add urgency to calls for more meaningful public transparency into the political advertising ecosystem.

cs.CY↗

arXiv2019

BLENDER: Enabling Local Search with a Hybrid Differential Privacy Model

We propose a hybrid model of differential privacy that considers a combination of regular and opt-in users who desire the differential privacy guarantees of the local privacy model and the trusted curator model, respectively. We demonstrate that within this model, it is possible to design a new type of blended algorithm for the task of privately computing the head of a search log. This blended approach provides significant improvements in the utility of obtained data compared to related work while providing users with their desired privacy guarantees. Specifically, on two large search click data sets, comprising 1.75 and 16 GB respectively, our approach attains NDCG values exceeding 95% across a range of privacy budget values.

cs.CR↗

arXiv2019

Discrimination through optimization: How Facebook's ad delivery can lead to skewed outcomes

The enormous financial success of online advertising platforms is partially due to the precise targeting features they offer. Although researchers and journalists have found many ways that advertisers can target---or exclude---particular groups of users seeing their ads, comparatively little attention has been paid to the implications of the platform's ad delivery process, comprised of the platform's choices about which users see which ads. It has been hypothesized that this process can "skew" ad delivery in ways that the advertisers do not intend, making some users less likely than others to see particular ads based on their demographic characteristics. In this paper, we demonstrate that such skewed delivery occurs on Facebook, due to market and financial optimization effects as well as the platform's own predictions about the "relevance" of ads to different groups of users. We find that both the advertiser's budget and the content of the ad each significantly contribute to the skew of Facebook's ad delivery. Critically, we observe significant skew in delivery along gender and racial lines for "real" ads for employment and housing opportunities despite neutral targeting parameters. Our results demonstrate previously unknown mechanisms that can lead to potentially discriminatory ad delivery, even when advertisers set their targeting parameters to be highly inclusive. This underscores the need for policymakers and platforms to carefully consider the role of the ad delivery optimization run by ad platforms themselves---and not just the targeting choices of advertisers---in preventing discrimination in digital advertising.

cs.CY↗

arXiv2019

Preference-Informed Fairness

We study notions of fairness in decision-making systems when individuals have diverse preferences over the possible outcomes of the decisions. Our starting point is the seminal work of Dwork et al. which introduced a notion of individual fairness (IF): given a task-specific similarity metric, every pair of individuals who are similarly qualified according to the metric should receive similar outcomes. We show that when individuals have diverse preferences over outcomes, requiring IF may unintentionally lead to less-preferred outcomes for the very individuals that IF aims to protect. A natural alternative to IF is the classic notion of fair division, envy-freeness (EF): no individual should prefer another individual's outcome over their own. Although EF allows for solutions where all individuals receive a highly-preferred outcome, EF may also be overly-restrictive. For instance, if many individuals agree on the best outcome, then if any individual receives this outcome, they all must receive it, regardless of each individual's underlying qualifications for the outcome. We introduce and study a new notion of preference-informed individual fairness (PIIF) that is a relaxation of both individual fairness and envy-freeness. At a high-level, PIIF requires that outcomes satisfy IF-style constraints, but allows for deviations provided they are in line with individuals' preferences. We show that PIIF can permit outcomes that are more favorable to individuals than any IF solution, while providing considerably more flexibility to the decision-maker than EF. In addition, we show how to efficiently optimize any convex objective over the outcomes subject to PIIF for a rich class of individual preferences. Finally, we demonstrate the broad applicability of the PIIF framework by extending our definitions and algorithms to the multiple-task targeted advertising setting introduced by Dwork and Ilvento.

cs.LG↗

arXiv2018

Differentially-Private "Draw and Discard" Machine Learning

In this work, we propose a novel framework for privacy-preserving client-distributed machine learning. It is motivated by the desire to achieve differential privacy guarantees in the local model of privacy in a way that satisfies all systems constraints using asynchronous client-server communication and provides attractive model learning properties. We call it "Draw and Discard" because it relies on random sampling of models for load distribution (scalability), which also provides additional server-side privacy protections and improved model quality through averaging. We present the mechanics of client and server components of "Draw and Discard" and demonstrate how the framework can be applied to learning Generalized Linear models. We then analyze the privacy guarantees provided by our approach against several types of adversaries and showcase experimental results that provide evidence for the framework's viability in practical deployments.

cs.CR↗