Julia Stoyanovich

Anna Richter, Julia Stoyanovich, Sebastian Schelter

Machine learning engineering (MLE) agents promise to automate end-to-end ML pipeline development from raw data and natural language instructions, potentially making ML accessible to non-technical domain experts. However, in sensitive and regulated domains, this abstraction creates a responsibility gap: end-users may lack visibility into design choices that affect correctness, robustness, fairness, and regulatory compliance. We argue that existing benchmarks are insufficient to assess whether MLE agents can be safely applied in such settings. We propose desiderata for a responsibility-centered evaluation framework and conduct an exploratory study on melanoma classification, focusing on fairness across skin tones as a responsibility constraint. When evaluating two recent MLE agents, we find that agent-generated pipelines show high variance and consistently underperform manually designed baselines in both predictive quality and fairness, despite fairness-oriented prompts. These preliminary results suggest that further research is needed towards redesigning MLE agents to allow humans to guide the search process and reliably assess the compliance and quality of the generated ML pipelines.

Andrew Bell, Oded Nov, Julia Stoyanovich

Increasingly, laws are being proposed and passed by governments around the world to regulate Artificial Intelligence (AI) systems implemented into the public and private sectors. Many of these regulations address the transparency of AI systems, and related citizen-aware issues like allowing individuals to have the right to an explanation about how an AI system makes a decision that impacts them. Yet, almost all AI governance documents to date have a significant drawback: they have focused on what to do (or what not to do) with respect to making AI systems transparent, but have left the brunt of the work to technologists to figure out how to build transparent systems. We fill this gap by proposing a novel stakeholder-first approach that assists technologists in designing transparent, regulatory compliant systems. We also describe a real-world case-study that illustrates how this approach can be used in practice.

Andrew Bell, Lucius Bynum, Nazarii Drushchak et al.

The ``impossibility theorem'' -- which is considered foundational in algorithmic fairness literature -- asserts that there must be trade-offs between common notions of fairness and performance when fitting statistical models, except in two special cases: when the prevalence of the outcome being predicted is equal across groups, or when a perfectly accurate predictor is used. However, theory does not always translate to practice. In this work, we challenge the implications of the impossibility theorem in practical settings. First, we show analytically that, by slightly relaxing the impossibility theorem (to accommodate a \textit{practitioner's} perspective of fairness), it becomes possible to identify a large set of models that satisfy seemingly incompatible fairness constraints. Second, we demonstrate the existence of these models through extensive experiments on five real-world datasets. We conclude by offering tools and guidance for practitioners to understand when -- and to what degree -- fairness along multiple criteria can be achieved. For example, if one allows only a small margin-of-error between metrics, there exists a large set of models simultaneously satisfying \emph{False Negative Rate Parity}, \emph{False Positive Rate Parity}, and \emph{Positive Predictive Value Parity}, even when there is a moderate prevalence difference between groups. This work has an important implication for the community: achieving fairness along multiple metrics for multiple groups (and their intersections) is much more possible than was previously believed.

Falaah Arif Khan, Eleni Manis, Julia Stoyanovich

In this work we use Equal Oppportunity (EO) doctrines from political philosophy to make explicit the normative judgements embedded in different conceptions of algorithmic fairness. We contrast formal EO approaches that narrowly focus on fair contests at discrete decision points, with substantive EO doctrines that look at people's fair life chances more holistically over the course of a lifetime. We use this taxonomy to provide a moral interpretation of the impossibility results as the incompatibility between different conceptions of a fair contest -- foward-looking versus backward-looking -- when people do not have fair life chances. We use this result to motivate substantive conceptions of algorithmic fairness and outline two plausible procedures based on the luck-egalitarian doctrine of EO, and Rawls's principle of fair equality of opportunity.

Joao Fonseca, Andrew Bell, Carlo Abrate et al.

Algorithmic systems are often called upon to assist in high-stakes decision making. In light of this, algorithmic recourse, the principle wherein individuals should be able to take action against an undesirable outcome made by an algorithmic system, is receiving growing attention. The bulk of the literature on algorithmic recourse to-date focuses primarily on how to provide recourse to a single individual, overlooking a critical element: the effects of a continuously changing context. Disregarding these effects on recourse is a significant oversight, since, in almost all cases, recourse consists of an individual making a first, unfavorable attempt, and then being given an opportunity to make one or several attempts at a later date - when the context might have changed. This can create false expectations, as initial recourse recommendations may become less reliable over time due to model drift and competition for access to the favorable outcome between individuals. In this work we propose an agent-based simulation framework for studying the effects of a continuously changing environment on algorithmic recourse. In particular, we identify two main effects that can alter the reliability of recourse for individuals represented by the agents: (1) competition with other agents acting upon recourse, and (2) competition with new agents entering the environment. Our findings highlight that only a small set of specific parameterizations result in algorithmic recourse that is reliable for agents over time. Consequently, we argue that substantial additional work is needed to understand recourse reliability over time, and to develop recourse methods that reward agents' effort.

Lucius E. J. Bynum, Joshua R. Loftus, Julia Stoyanovich

Counterfactuals are often described as 'retrospective,' focusing on hypothetical alternatives to a realized past. This description relates to an often implicit assumption about the structure and stability of exogenous variables in the system being modeled -- an assumption that is reasonable in many settings where counterfactuals are used. In this work, we consider cases where we might reasonably make a different assumption about exogenous variables, namely, that the exogenous noise terms of each unit do exhibit some unit-specific structure and/or stability. This leads us to a different use of counterfactuals -- a 'forward-looking' rather than 'retrospective' counterfactual. We introduce "counterfactual treatment choice," a type of treatment choice problem that motivates using forward-looking counterfactuals. We then explore how mismatches between interventional versus forward-looking counterfactual approaches to treatment choice, consistent with different assumptions about exogenous noise, can lead to counterintuitive results.

Lucas Rosenblatt, Joshua Allen, Julia Stoyanovich

Differentially private (DP) synthetic data generation is a practical method for improving access to data as a means to encourage productive partnerships. One issue inherent to DP is that the "privacy budget" is generally "spent" evenly across features in the data set. This leads to good statistical parity with the real data, but can undervalue the conditional probabilities and marginals that are critical for predictive quality of synthetic data. Further, loss of predictive quality may be non-uniform across the data set, with subsets that correspond to minority groups potentially suffering a higher loss. In this paper, we develop ensemble methods that distribute the privacy budget "wisely" to maximize predictive accuracy of models trained on DP data, and "fairly" to bound potential disparities in accuracy across groups and reduce inequality. Our methods are based on the insights that feature importance can inform how privacy budget is allocated, and, further, that per-group feature importance and fairness-related performance objectives can be incorporated in the allocation. These insights make our methods tunable to social contexts, allowing data owners to produce balanced synthetic data for predictive analysis.

Falaah Arif Khan, Julia Stoyanovich

In this paper we revisit the bias-variance decomposition of model error from the perspective of designing a fair classifier: we are motivated by the widely held socio-technical belief that noise variance in large datasets in social domains tracks demographic characteristics such as gender, race, disability, etc. We propose a conditional-iid (ciid) model built from group-specific classifiers that seeks to improve on the trade-offs made by a single model (iid setting). We theoretically analyze the bias-variance decomposition of different models in the Gaussian Mixture Model, and then empirically test our setup on the COMPAS and folktables datasets. We instantiate the ciid model with two procedures that improve "fairness" by conditioning out undesirable effects: first, by conditioning directly on sensitive attributes, and second, by clustering samples into groups and conditioning on cluster membership (blind to protected group membership). Our analysis suggests that there might be principled procedures and concrete real-world use cases under which conditional models are preferred, and our striking empirical results strongly indicate that non-iid settings, such as the ciid setting proposed here, might be more suitable for big data applications in social contexts.

Falaah Arif Khan, Denys Herasymuk, Nazar Protsiv et al.

Recent research reveals that machine learning (ML) models are highly sensitive to minor changes in their training procedure, such as the inclusion or exclusion of a single data point, leading to conflicting predictions on individual data points; a property termed as arbitrariness or instability in ML pipelines in prior work. Drawing from the uncertainty literature, we show that stability decomposes into epistemic and aleatoric components, capturing the consistency and confidence in prediction, respectively. We use this decomposition to provide two main contributions. Our first contribution is an extensive empirical evaluation. We find that (i) epistemic instability can be reduced with more training data whereas aleatoric instability cannot; (ii) state-of-the-art ML models have aleatoric instability as high as 79% and aleatoric instability disparities among demographic groups as high as 29% in popular fairness benchmarks; and (iii) fairness pre-processing interventions generally increase aleatoric instability more than in-processing interventions, and both epistemic and aleatoric instability are highly sensitive to data-processing interventions and model architecture. Our second contribution is a practical solution to the problem of systematic arbitrariness. We propose a model selection procedure that includes epistemic and aleatoric criteria alongside existing accuracy and fairness criteria, and show that it successfully narrows down a large set of good models (50-100 on our datasets) to a handful of stable, fair and accurate ones. We built and publicly released a python library to measure epistemic and aleatoric multiplicity in any ML pipeline alongside existing confusion-matrix-based metrics, providing practitioners with a rich suite of evaluation metrics to use to define a more precise criterion during model selection.

Joao Fonseca, Julia Stoyanovich

Computing the importance of features in supervised classification tasks is critical for model interpretability. Shapley values are a widely used approach for explaining model predictions, but require direct access to the underlying model, an assumption frequently violated in real-world deployments. Further, even when model access is possible, their exact computation may be prohibitively expensive. We investigate whether meaningful Shapley value estimations can be obtained in a zero-shot setting, using only the input data distribution and no evaluations of the target model. To this end, we introduce ExplainerPFN, a tabular foundation model built on TabPFN that is pretrained on synthetic datasets generated from random structural causal models and supervised using exact or near-exact Shapley values. Once trained, ExplainerPFN predicts feature attributions for unseen tabular datasets without model access, gradients, or example explanations. Our contributions are fourfold: (1) we show that few-shot learning-based explanations can achieve high fidelity to SHAP values with as few as two reference observations; (2) we propose ExplainerPFN, the first zero-shot method for estimating Shapley values without access to the underlying model or reference explanations; (3) we provide an open-source implementation of ExplainerPFN, including the full training pipeline and synthetic data generator; and (4) through extensive experiments on real and synthetic datasets, we show that ExplainerPFN achieves performance competitive with few-shot surrogate explainers that rely on 2-10 SHAP examples.

Rima Hazra, Bikram Ghuku, Ilona Marchenko et al.

Large language models are rapidly being deployed as AI tutors, yet current evaluation paradigms assess problem-solving accuracy and generic safety in isolation, failing to capture whether a model is simultaneously pedagogically effective and safe across student-tutor interaction. We argue that tutoring safety is fundamentally different from conventional LLM safety: the primary risk is not toxic content but the quiet erosion of learning through answer over-disclosure, misconception reinforcement, and the abdication of scaffolding. To systematically study this failure mode, we introduce SafeTutors, a benchmark that jointly evaluates safety and pedagogy across mathematics, physics, and chemistry. SafeTutors is organized around a theoretically grounded risk taxonomy comprising 11 harm dimensions and 48 sub-risks drawn from learning-science literature. We uncover that all models show broad harm; scale doesn't reliably help; and multi-turn dialogue worsens behavior, with pedagogical failures rising from 17.7% to 77.8%. Harms also vary by subject, so mitigations must be discipline-aware, and single-turn "safe/helpful" results can mask systematic tutor failure over extended interaction.

Joao Fonseca, Andrew Bell, Julia Stoyanovich

Large Language Models (LLMs) have been shown to be susceptible to jailbreak attacks, or adversarial attacks used to illicit high risk behavior from a model. Jailbreaks have been exploited by cybercriminals and blackhat actors to cause significant harm, highlighting the critical need to safeguard widely-deployed models. Safeguarding approaches, which include fine-tuning models or having LLMs "self-reflect", may lengthen the inference time of a model, incur a computational penalty, reduce the semantic fluency of an output, and restrict ``normal'' model behavior. Importantly, these Safety-Performance Trade-offs (SPTs) remain an understudied area. In this work, we introduce a novel safeguard, called SafeNudge, that combines Controlled Text Generation with "nudging", or using text interventions to change the behavior of a model. SafeNudge triggers during text-generation while a jailbreak attack is being executed, and can reduce successful jailbreak attempts by 30% by guiding the LLM towards a safe responses. It adds minimal latency to inference and has a negligible impact on the semantic fluency of outputs. Further, we allow for tunable SPTs. SafeNudge is open-source and available through https://pypi.org/, and is compatible with models loaded with the Hugging Face "transformers" library.

Falaah Arif Khan, Denys Herasymuk, Nazar Protsiv et al.

Data missingness is a practical challenge of sustained interest to the scientific community. In this paper, we present Shades-of-Null, an evaluation suite for responsible missing value imputation. Our work is novel in two ways (i) we model realistic and socially-salient missingness scenarios that go beyond Rubin's classic Missing Completely at Random (MCAR), Missing At Random (MAR) and Missing Not At Random (MNAR) settings, to include multi-mechanism missingness (when different missingness patterns co-exist in the data) and missingness shift (when the missingness mechanism changes between training and test) (ii) we evaluate imputers holistically, based on imputation quality and imputation fairness, as well as on the predictive performance, fairness and stability of the models that are trained and tested on the data post-imputation. We use Shades-of-Null to conduct a large-scale empirical study involving 29,736 experimental pipelines, and find that while there is no single best-performing imputation approach for all missingness types, interesting trade-offs arise between predictive performance, fairness and stability, based on the combination of missingness scenario, imputer choice, and the architecture of the predictive model. We make Shades-of-Null publicly available, to enable researchers to rigorously evaluate missing value imputation methods on a wide range of metrics in plausible and socially meaningful scenarios.

Andrew Bell, Julia Stoyanovich

Concerns about the risks and harms posed by artificial intelligence (AI) have resulted in significant study into algorithmic transparency, giving rise to a sub-field known as Explainable AI (XAI). Unfortunately, despite a decade of development in XAI, an existential challenge remains: progress in research has not been fully translated into the actual implementation of algorithmic transparency by organizations. In this work, we test an approach for addressing the challenge by creating transparency advocates, or motivated individuals within organizations who drive a ground-up cultural shift towards improved algorithmic transparency. Over several years, we created an open-source educational workshop on algorithmic transparency and advocacy. We delivered the workshop to professionals across two separate domains to improve their algorithmic transparency literacy and willingness to advocate for change. In the weeks following the workshop, participants applied what they learned, such as speaking up for algorithmic transparency at an organization-wide AI strategy meeting. We also make two broader observations: first, advocacy is not a monolith and can be broken down into different levels. Second, individuals' willingness for advocacy is affected by their professional field. For example, news and media professionals may be more likely to advocate for algorithmic transparency than those working at technology start-ups.

Alene K. Rhea, Kelsey Markey, Lauren D'Arinzo et al.

Automated hiring systems are among the fastest-developing of all high-stakes AI systems. Among these are algorithmic personality tests that use insights from psychometric testing, and promise to surface personality traits indicative of future success based on job seekers' resumes or social media profiles. We interrogate the validity of such systems using stability of the outputs they produce, noting that reliability is a necessary, but not a sufficient, condition for validity. Our approach is to (a) develop a methodology for an external audit of stability of predictions made by algorithmic personality tests, and (b) instantiate this methodology in an audit of two systems, Humantic AI and Crystal. Crucially, rather than challenging or affirming the assumptions made in psychometric testing -- that personality is a meaningful and measurable construct, and that personality traits are indicative of future success on the job -- we frame our methodology around testing the underlying assumptions made by the vendors of the algorithmic personality tests themselves. Our main contribution is the development of a socio-technical framework for auditing the stability of algorithmic systems. This contribution is supplemented with an open-source software library that implements the technical components of the audit, and can be used to conduct similar stability audits of algorithmic systems. We instantiate our framework with the audit of two real-world personality prediction systems, namely Humantic AI and Crystal. The application of our audit framework demonstrates that both these systems show substantial instability with respect to key facets of measurement, and hence cannot be considered valid testing instruments.

Venetia Pliatsika, Joao Fonseca, Kateryna Akhynko et al.

Algorithmic decisions in critical domains such as hiring, college admissions, and lending are often based on rankings. Given the impact of these decisions on individuals, organizations, and population groups, it is essential to understand them - to help individuals improve their ranking position, design better ranking procedures, and ensure legal compliance. In this paper, we argue that explainability methods for classification and regression, such as SHAP, are insufficient for ranking tasks, and present ShaRP - Shapley Values for Rankings and Preferences - a framework that explains the contributions of features to various aspects of a ranked outcome. ShaRP computes feature contributions for various ranking-specific profit functions, such as rank and top-k, and also includes a novel Shapley value-based method for explaining pairwise preference outcomes. We provide a flexible implementation of ShaRP, capable of efficiently and comprehensively explaining ranked and pairwise outcomes over tabular data, in score-based ranking and learning-to-rank tasks. Finally, we develop a comprehensive evaluation methodology for ranking explainability methods, showing through qualitative, quantitative, and usability studies that our rank-aware QoIs offer complementary insights, scale effectively, and help users interpret ranked outcomes in practice.

Hyunseung Hwang, Andrew Bell, Joao Fonseca et al.

Local feature-based explanations are a key component of the XAI toolkit. These explanations compute feature importance values relative to an ``interpretable'' feature representation. In tabular data, feature values themselves are often considered interpretable. This paper examines the impact of data engineering choices on local feature-based explanations. We demonstrate that simple, common data engineering techniques, such as representing age with a histogram or encoding race in a specific way, can manipulate feature importance as determined by popular methods like SHAP. Notably, the sensitivity of explanations to feature representation can be exploited by adversaries to obscure issues like discrimination. While the intuition behind these results is straightforward, their systematic exploration has been lacking. Previous work has focused on adversarial attacks on feature-based explainers by biasing data or manipulating models. To the best of our knowledge, this is the first study demonstrating that explainers can be misled by standard, seemingly innocuous data engineering techniques.

Andrew Bell, Joao Fonseca, Carlo Abrate et al.

Algorithmic recourse -- providing recommendations to those affected negatively by the outcome of an algorithmic system on how they can take action and change that outcome -- has gained attention as a means of giving persons agency in their interactions with artificial intelligence (AI) systems. Recent work has shown that even if an AI decision-making classifier is ``fair'' (according to some reasonable criteria), recourse itself may be unfair due to differences in the initial circumstances of individuals, compounding disparities for marginalized populations and requiring them to exert more effort than others. There is a need to define more methods and metrics for evaluating fairness in recourse that span a range of normative views of the world, and specifically those that take into account time. Time is a critical element in recourse because the longer it takes an individual to act, the more the setting may change due to model or data drift. This paper seeks to close this research gap by proposing two notions of fairness in recourse that are in normative alignment with substantive equality of opportunity, and that consider time. The first considers the (often repeated) effort individuals exert per successful recourse event, and the second considers time per successful recourse event. Building upon an agent-based framework for simulating recourse, this paper demonstrates how much effort is needed to overcome disparities in initial circumstances. We then proposes an intervention to improve the fairness of recourse by rewarding effort, and compare it to existing strategies.

Lucas Rosenblatt, Julia Stoyanovich, Christopher Musco

Differentially private (DP) mechanisms have been deployed in a variety of high-impact social settings (perhaps most notably by the U.S. Census). Since all DP mechanisms involve adding noise to results of statistical queries, they are expected to impact our ability to accurately analyze and learn from data, in effect trading off privacy with utility. Alarmingly, the impact of DP on utility can vary significantly among different sub-populations. A simple way to reduce this disparity is with stratification. First compute an independent private estimate for each group in the data set (which may be the intersection of several protected classes), then, to compute estimates of global statistics, appropriately recombine these group estimates. Our main observation is that naive stratification often yields high-accuracy estimates of population-level statistics, without the need for additional privacy budget. We support this observation theoretically and empirically. Our theoretical results center on the private mean estimation problem, while our empirical results center on extensive experiments on private data synthesis to demonstrate the effectiveness of stratification on a variety of private mechanisms. Overall, we argue that this straightforward approach provides a strong baseline against which future work on reducing utility disparities of DP mechanisms should be compared.

Shlomi Hod, Lucas Rosenblatt, Julia Stoyanovich

Differentially private (DP) machine learning often relies on the availability of public data for tasks like privacy-utility trade-off estimation, hyperparameter tuning, and pretraining. While public data assumptions may be reasonable in text and image domains, they are less likely to hold for tabular data due to tabular data heterogeneity across domains. We propose leveraging powerful priors to address this limitation; specifically, we synthesize realistic tabular data directly from schema-level specifications - such as variable names, types, and permissible ranges - without ever accessing sensitive records. To that end, this work introduces the notion of "surrogate" public data - datasets generated independently of sensitive data, which consume no privacy loss budget and are constructed solely from publicly available schema or metadata. Surrogate public data are intended to encode plausible statistical assumptions (informed by publicly available information) into a dataset with many downstream uses in private mechanisms. We automate the process of generating surrogate public data with large language models (LLMs); in particular, we propose two methods: direct record generation as CSV files, and automated structural causal model (SCM) construction for sampling records. Through extensive experiments, we demonstrate that surrogate public tabular data can effectively replace traditional public data when pretraining differentially private tabular classifiers. To a lesser extent, surrogate public data are also useful for hyperparameter tuning of DP synthetic data generators, and for estimating the privacy-utility tradeoff.

Caterina Fuligni, Daniel Dominguez Figaredo, Julia Stoyanovich

In recent years, Large Language Models (LLMs) rapidly gained popularity across all parts of society, including education. After initial skepticism and bans, many schools have chosen to embrace this new technology by integrating it into their curricula in the form of virtual tutors and teaching assistants. However, neither the companies developing this technology nor the public institutions involved in its implementation have set up a formal system to collect feedback from the stakeholders impacted by them. In this paper, we argue that understanding the perceptions of those directly or indirectly impacted by LLMs in the classroom, including parents and school staff, is essential for ensuring responsible use of AI in this critical domain. Our contributions are two-fold. First, we propose the Contextualized Perceptions for the Adoption of LLMs in Education (Co-PALE) framework, which can be used to systematically elicit perceptions and inform whether and how LLM-based tools should be designed, developed, and deployed in the classroom. Second, we explain how our framework can be used to ground specific rubrics for eliciting perceptions of the relevant stakeholders in view of specific goals and context of implementation. Overall, Co-PALE is a practical step toward helping educational agents, policymakers, researchers, and technologists ensure the responsible and effective deployment of LLM-based systems across diverse learning contexts.

Hyunseung Hwang, Seungeun Lee, Lucas Rosenblatt et al.

Post-hoc explanations are widely used to justify, contest, and audit automated decisions in high-stakes domains. SHAP, in particular, is often treated as a reliable account of which features drove an individual prediction. Yet SHAP explanations can vary substantially across repeated runs even when the input, task, and trained model are held fixed. We term this phenomenon explanation multiplicity: multiple internally valid but substantively different explanations for the same decision. We present a methodology to characterize multiplicity in feature-attribution explanations and to disentangle sources due to model training/selection from stochasticity intrinsic to the explanation pipeline. We further show that apparent stability depends on the metric: magnitude-based distances can remain near zero while rank-based measures reveal substantial churn in the identity and ordering of top features. To contextualize observed disagreement, we derive randomized baseline values under plausible null models. Across datasets, model classes, and confidence regimes, we find explanation multiplicity is pervasive and persists even for high-confidence predictions, highlighting the need for metrics and baselines that match the intended use of explanations.

Viktoriia Makovska, George Fletcher, Julia Stoyanovich

We present ONION, a multi-layered framework for participatory Entity-Relationship (ER) modeling that integrates insights from design justice, participatory AI, and conceptual modeling. ONION introduces a five-stage methodology: Observe, Nurture, Integrate, Optimize, Normalize. It supports progressive abstraction from unstructured stakeholder input to structured ER diagrams. Our approach aims to reduce designer bias, promote inclusive participation, and increase transparency through the modeling process. We evaluate ONION through real-world workshops focused on sociotechnical systems in Ukraine, highlighting how diverse stakeholder engagement leads to richer data models and deeper mutual understanding. Early results demonstrate ONION's potential to host diversity in early-stage data modeling. We conclude with lessons learned, limitations and challenges involved in scaling and refining the framework for broader adoption.

Denys Herasymuk, Nazar Protsiv, Julia Stoyanovich

Developing machine learning (ML) models requires a deep understanding of real-world problems, which are inherently multi-objective. In this paper, we present VirnyFlow, the first design space for responsible model development, designed to assist data scientists in building ML pipelines that are tailored to the specific context of their problem. Unlike conventional AutoML frameworks, VirnyFlow enables users to define customized optimization criteria, perform comprehensive experimentation across pipeline stages, and iteratively refine models in alignment with real-world constraints. Our system integrates evaluation protocol definition, multi-objective Bayesian optimization, cost-aware multi-armed bandits, query optimization, and distributed parallelism into a unified architecture. We show that VirnyFlow significantly outperforms state-of-the-art AutoML systems in both optimization quality and scalability across five real-world benchmarks, offering a flexible, efficient, and responsible alternative to black-box automation in ML development.

George Fletcher, Olha Nahurna, Matvii Prytula et al.

Data models are necessary for the birth of data and of any data-driven system. Indeed, every algorithm, every machine learning model, every statistical model, and every database has an underlying data model without which the system would not be usable. Hence, data models are excellent sites for interrogating the (material, social, political, ...) conditions giving rise to a data system. Towards this, drawing inspiration from literary criticism, we propose to closely read data models in the same spirit as we closely read literary artifacts. Close readings of data models reconnect us with, among other things, the materiality, the genealogies, the techne, the closed nature, and the design of technical systems. While recognizing from literary theory that there is no one correct way to read, it is nonetheless critical to have systematic guidance for those unfamiliar with close readings. This is especially true for those trained in the computing and data sciences, who too often are enculturated to set aside the socio-political aspects of data work. A systematic methodology for reading data models currently does not exist. To fill this gap, we present the CREDAL methodology for close readings of data models. We detail our iterative development process and present results of a qualitative evaluation of CREDAL demonstrating its usability, usefulness, and effectiveness in the critical study of data.

Lucius E. J. Bynum, Joshua R. Loftus, Julia Stoyanovich

Counterfactuals and counterfactual reasoning underpin numerous techniques for auditing and understanding artificial intelligence (AI) systems. The traditional paradigm for counterfactual reasoning in this literature is the interventional counterfactual, where hypothetical interventions are imagined and simulated. For this reason, the starting point for causal reasoning about legal protections and demographic data in AI is an imagined intervention on a legally-protected characteristic, such as ethnicity, race, gender, disability, age, etc. We ask, for example, what would have happened had your race been different? An inherent limitation of this paradigm is that some demographic interventions -- like interventions on race -- may not translate into the formalisms of interventional counterfactuals. In this work, we explore a new paradigm based instead on the backtracking counterfactual, where rather than imagine hypothetical interventions on legally-protected characteristics, we imagine alternate initial conditions while holding these characteristics fixed. We ask instead, what would explain a counterfactual outcome for you as you actually are or could be? This alternate framework allows us to address many of the same social concerns, but to do so while asking fundamentally different questions that do not rely on demographic interventions.

Lucius E. J. Bynum, Joshua R. Loftus, Julia Stoyanovich

A significant body of research in the data sciences considers unfair discrimination against social categories such as race or gender that could occur or be amplified as a result of algorithmic decisions. Simultaneously, real-world disparities continue to exist, even before algorithmic decisions are made. In this work, we draw on insights from the social sciences brought into the realm of causal modeling and constrained optimization, and develop a novel algorithmic framework for tackling pre-existing real-world disparities. The purpose of our framework, which we call the "impact remediation framework," is to measure real-world disparities and discover the optimal intervention policies that could help improve equity or access to opportunity for those who are underserved with respect to an outcome of interest. We develop a disaggregated approach to tackling pre-existing disparities that relaxes the typical set of assumptions required for the use of social categories in structural causal models. Our approach flexibly incorporates counterfactuals and is compatible with various ontological assumptions about the nature of social categories. We demonstrate impact remediation with a hypothetical case study and compare our disaggregated approach to an existing state-of-the-art approach, comparing its structure and resulting policy recommendations. In contrast to most work on optimal policy learning, we explore disparity reduction itself as an objective, explicitly focusing the power of algorithms on reducing inequality.

Falaah Arif Khan, Eleni Manis, Julia Stoyanovich

Recent interest in codifying fairness in Automated Decision Systems (ADS) has resulted in a wide range of formulations of what it means for an algorithmic system to be fair. Most of these propositions are inspired by, but inadequately grounded in, political philosophy scholarship. This paper aims to correct that deficit. We introduce a taxonomy of fairness ideals using doctrines of Equality of Opportunity (EOP) from political philosophy, clarifying their conceptions in philosophy and the proposed codification in fair machine learning. We arrange these fairness ideals onto an EOP spectrum, which serves as a useful frame to guide the design of a fair ADS in a given context. We use our fairness-as-EOP framework to re-interpret the impossibility results from a philosophical perspective, as the in-compatibility between different value systems, and demonstrate the utility of the framework with several real-world and hypothetical examples. Through our EOP-framework we hope to answer what it means for an ADS to be fair from a moral and political philosophy standpoint, and to pave the way for similar scholarship from ethics and legal experts.

Meike Zehlike, Ke Yang, Julia Stoyanovich

In the past few years, there has been much work on incorporating fairness requirements into algorithmic rankers, with contributions coming from the data management, algorithms, information retrieval, and recommender systems communities. In this survey we give a systematic overview of this work, offering a broad perspective that connects formalizations and algorithmic approaches across subfields. An important contribution of our work is in developing a common narrative around the value frameworks that motivate specific fairness-enhancing interventions in ranking. This allows us to unify the presentation of mitigation objectives and of algorithmic techniques to help meet those objectives or identify trade-offs. In this survey, we describe four classification frameworks for fairness-enhancing interventions, along which we relate the technical methods surveyed in this paper, discuss evaluation datasets, and present technical work on fairness in score-based ranking. Then, we present methods that incorporate fairness in supervised learning, and also give representative examples of recent work on fairness in recommendation and matchmaking systems. We also discuss evaluation frameworks for fair score-based ranking and fair learning-to-rank, and draw a set of recommendations for the evaluation of fair ranking methods.

Ke Yang, Joshua R. Loftus, Julia Stoyanovich

In this paper we propose a causal modeling approach to intersectional fairness, and a flexible, task-specific method for computing intersectionally fair rankings. Rankings are used in many contexts, ranging from Web search results to college admissions, but causal inference for fair rankings has received limited attention. Additionally, the growing literature on causal fairness has directed little attention to intersectionality. By bringing these issues together in a formal causal framework we make the application of intersectionality in fair machine learning explicit, connected to important real world effects and domain knowledge, and transparent about technical limitations. We experimentally evaluate our approach on real and synthetic datasets, exploring its behaviour under different structural assumptions.

Julia Stoyanovich, Armanda Lewis

Although numerous ethics courses are available, with many focusing specifically on technology and computer ethics, pedagogical approaches employed in these courses rely exclusively on texts rather than on software development or data analysis. Technical students often consider these courses unimportant and a distraction from the "real" material. To develop instructional materials and methodologies that are thoughtful and engaging, we must strive for balance: between texts and coding, between critique and solution, and between cutting-edge research and practical applicability. Finding such balance is particularly difficult in the nascent field of responsible data science (RDS), where we are only starting to understand how to interface between the intrinsically different methodologies of engineering and social sciences. In this paper we recount a recent experience in developing and teaching an RDS course to graduate and advanced undergraduate students in data science. We then dive into an area that is critically important to RDS -- transparency and interpretability of machine-assisted decision-making, and tie this area to the needs of emerging RDS curricula. Recounting our own experience, and leveraging literature on pedagogical methods in data science and beyond, we propose the notion of an "object-to-interpret-with". We link this notion to "nutritional labels" -- a family of interpretability tools that are gaining popularity in RDS research and practice. With this work we aim to contribute to the nascent area of RDS education, and to inspire others in the community to come together to develop a deeper theoretical understanding of the pedagogical needs of RDS, and contribute concrete educational materials and methodologies that others can use. All course materials are publicly available at https://dataresponsibly.github.io/courses.

Sebastian Schelter, Yuxuan He, Jatin Khilnani et al.

The importance of incorporating ethics and legal compliance into machine-assisted decision-making is broadly recognized. Further, several lines of recent work have argued that critical opportunities for improving data quality and representativeness, controlling for bias, and allowing humans to oversee and impact computational processes are missed if we do not consider the lifecycle stages upstream from model training and deployment. Yet, very little has been done to date to provide system-level support to data scientists who wish to develop and deploy responsible machine learning methods. We aim to fill this gap and present FairPrep, a design and evaluation framework for fairness-enhancing interventions. FairPrep is based on a developer-centered design, and helps data scientists follow best practices in software engineering and machine learning. As part of our contribution, we identify shortcomings in existing empirical studies for analyzing fairness-enhancing interventions. We then show how FairPrep can be used to measure the impact of sound best practices, such as hyperparameter tuning and feature scaling. In particular, our results suggest that the high variability of the outcomes of fairness-enhancing interventions observed in previous studies is often an artifact of a lack of hyperparameter tuning. Further, we show that the choice of a data cleaning method can impact the effectiveness of fairness-enhancing interventions.

Ke Yang, Vasilis Gkatzelis, Julia Stoyanovich

Many set selection and ranking algorithms have recently been enhanced with diversity constraints that aim to explicitly increase representation of historically disadvantaged populations, or to improve the overall representativeness of the selected set. An unintended consequence of these constraints, however, is reduced in-group fairness: the selected candidates from a given group may not be the best ones, and this unfairness may not be well-balanced across groups. In this paper we study this phenomenon using datasets that comprise multiple sensitive attributes. We then introduce additional constraints, aimed at balancing the \in-group fairness across groups, and formalize the induced optimization problems as integer linear programs. Using these programs, we conduct an experimental evaluation with real datasets, and quantify the feasible trade-offs between balance and overall performance in the presence of diversity constraints.

Benny Kimelfeld, Phokion G. Kolaitis, Julia Stoyanovich

We develop a novel framework that aims to create bridges between the computational social choice and the database management communities. This framework enriches the tasks currently supported in computational social choice with relational database context, thus making it possible to formulate sophisticated queries about voting rules, candidates, voters, issues, and positions. At the conceptual level, we give rigorous semantics to queries in this framework by introducing the notions of necessary answers and possible answers to queries. At the technical level, we embark on an investigation of the computational complexity of the necessary answers. We establish a number of results about the complexity of the necessary answers of conjunctive queries involving positional scoring rules that contrast sharply with earlier results about the complexity of the necessary winners.

Ke Yang, Julia Stoyanovich, Abolfazl Asudeh et al.

Algorithmic decisions often result in scoring and ranking individuals to determine credit worthiness, qualifications for college admissions and employment, and compatibility as dating partners. While automatic and seemingly objective, ranking algorithms can discriminate against individuals and protected groups, and exhibit low diversity. Furthermore, ranked results are often unstable --- small changes in the input data or in the ranking methodology may lead to drastic changes in the output, making the result uninformative and easy to manipulate. Similar concerns apply in cases where items other than individuals are ranked, including colleges, academic departments, or products. In this demonstration we present Ranking Facts, a Web-based application that generates a "nutritional label" for rankings. Ranking Facts is made up of a collection of visual widgets that implement our latest research results on fairness, stability, and transparency for rankings, and that communicate details of the ranking methodology, or of the output, to the end user. We will showcase Ranking Facts on real datasets from different domains, including college rankings, criminal risk assessment, and financial services.