Despite achieving promising fairness-error trade-offs, in-processing mitigation techniques for group fairness cannot be employed in numerous practical applications with limited computation resources or no access to the training pipeline of the prediction model. In these situations, post-processing is a viable alternative. However, current methods are tailored to specific problem settings and fairness definitions and hence, are not as broadly applicable as in-processing. In this work, we propose a framework that turns any regularized in-processing method into a post-processing approach. This procedure prescribes a way to obtain post-processing techniques for a much broader range of problem settings than the prior post-processing literature. We show theoretically and through extensive experiments that our framework preserves the good fairness-error trade-offs achieved with in-processing and can improve over the effectiveness of prior post-processing methods. Finally, we demonstrate several advantages of a modular mitigation strategy that disentangles the training of the prediction model from the fairness mitigation, including better performance on tasks with partial group labels.

Preethi Lahoti, Nicholas Blumm, Xiao Ma, Raghavendra Kotikalapudi, Sahitya Potluri, Qijun Tan, Hansa Srinivasan, Ben Packer, Ahmad Beirami, Alex Beutel, Jilin Chen. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. 2023.

Abstract Reliably predicting potential failure risks of machine learning (ML) systems when deployed with production data is a crucial aspect of trustworthy AI. This paper introduces the Risk Advisor, a novel post-hoc meta-learner for estimating failure risks and predictive uncertainties of any already-trained black-box classification model. In addition to providing a risk score, the Risk Advisor decomposes the uncertainty estimates into aleatoric and epistemic uncertainty components, thus giving informative insights into the sources of uncertainty inducing the failures. Consequently, Risk Advisor can distinguish between failures caused by data variability, data shifts and model limitations and provide useful guidance on appropriate risk mitigation actions (e.g., collecting more data to counter data shift). Extensive experiments on various families of black-box classification models and on real-world and synthetic datasets covering common ML failure scenarios show that the Risk Advisor reliably predicts deployment-time failure risks in all the scenarios, and outperforms strong baselines.

Traditional approaches to ensure group fairness in algorithmic decision making aim to equalize "total" error rates for different subgroups in the population. In contrast, we argue that the fairness approaches should instead focus only on equalizing errors arising due to model uncertainty (a.k.a epistemic uncertainty), caused due to lack of knowledge about the best model or due to lack of data. In other words, our proposal calls for ignoring the errors that occur due to uncertainty inherent in the data, i.e., aleatoric uncertainty. We draw a connection between predictive multiplicity and model uncertainty and argue that the techniques from predictive multiplicity could be used to identify errors made due to model uncertainty. We propose scalable convex proxies to come up with classifiers that exhibit predictive multiplicity and empirically show that our methods are comparable in performance and up to four orders of magnitude faster than the current state-of-the-art. We further pro- pose methods to achieve our goal of equalizing group error rates arising due to model uncertainty in algorithmic decision making and demonstrate the effectiveness of these methods using synthetic and real-world datasets

Reliably predicting potential failure risks of machine learning (ML) systems when deployed with production data is a crucial aspect of trustworthy AI. This paper introduces Risk Advisor, a novel post-hoc meta-learner for estimating failure risks and predictive uncertainties of any already-trained black-box classification model. In addition to providing a risk score, the Risk Advisor decomposes the uncertainty estimates into aleatoric and epistemic uncertainty components, thus giving informative insights into the sources of uncertainty inducing the failures. Consequently, Risk Advisor can distinguish between failures caused by data variability, data shifts and model limitations and advise on mitigation actions (e.g., collecting more data to counter data shift). Extensive experiments on real-world datasets covering a variety of ML failure scenarios show that the Risk Advisor reliably predicts deployment-time failure risks in all the scenarios, and outperforms strong baselines.

Much of the previous machine learning (ML) fairness literature assumes that protected features such as race and sex are present in the dataset, and relies upon them to mitigate fairness concerns. However, in practice factors like privacy and regulation often preclude the collection of protected features, or their use for training or inference, severely limiting the applicability of traditional fairness research. Therefore, we ask: How can we train a ML model to improve fairness when we do not even know the protected group memberships? In this work we address this problem by proposing Adversarially Reweighted Learning (ARL). In particular, we hypothesize that non-protected features and task labels are valuable for identifying fairness issues, and can be used to co-train an adversarial reweighting approach for improving fairness. Our results show that ARL improves Rawlsian Max-Min fairness, with notable AUC improvements for worst-case protected groups in multiple datasets, outperforming state-of-the-art alternatives.

People are rated and ranked, towards algorithmic decision making in an increasing number of applications, typically based on machine learning. Research on how to incorporate fairness into such tasks has prevalently pursued the paradigm of group fairness: giving adequate success rates to specifically protected groups. In contrast, the alternative paradigm of individual fairness has received relatively little attention, and this paper advances this less explored direction. The paper introduces a method for probabilistically mapping user records into a low-rank representation that reconciles individual fairness and the utility of classifiers and rankings in downstream applications. Our notion of individual fairness requires that users who are similar in all task-relevant attributes such as job qualification, and disregarding all potentially discriminating attributes such as gender, should have similar outcomes. We demonstrate the versatility of our method by applying it to classification and learning-to-rank tasks on a variety of real-world datasets. Our experiments show substantial improvements over the best prior work for this setting.

We revisit the notion of individual fairness proposed by Dwork et al. A central challenge in operationalizing their approach is the difficulty in eliciting a human specification of a similarity metric. In this paper, we propose an operationalization of individual fairness that does not rely on a human specification of a distance metric. Instead, we propose novel approaches to elicit and leverage side-information on equally deserving individuals to counter subordination between social groups. We model this knowledge as a fairness graph, and learn a unified Pairwise Fair Representation (PFR) of the data that captures both data-driven similarity between individuals and the pairwise side-information in fairness graph. We elicit fairness judgments from a variety of sources, including human judgments for two real-world datasets on recidivism prediction (COMPAS) and violent neighborhood prediction (Crime & Communities). Our experiments show that the PFR model for operationalizing individual fairness is practically viable.

People are shifting from traditional news sources to online news at an incredibly fast rate. However, the technology behind online news consumption promotes content that confirms the users» existing point of view. This phenomenon has led to polarization of opinions and intolerance towards opposing views. Thus, a key problem is to model information filter bubbles on social media and design methods to eliminate them. In this paper, we use a machine-learning approach to learn a liberal-conservative ideology space on Twitter, and show how we can use the learned latent space to tackle the filter bubble problem.

Set intersection counting appears as a subroutine in many techniques used in natural language processing, in which similarity is often measured as a function of document cooccurence counts between pairs of noun phrases or entities. Such techniques include clustering of text phrases and named entities, topic labeling, entity disambiguation, sentiment analysis, and search for synonyms. These techniques can have real-time constraints that require very fast computation of thousands of set intersection counting queries with little space overhead and minimal error. On one hand, while sketching techniques for approximate intersection counting exist and have very fast query time, many have issues with accuracy, especially for pairs of lists that have low Jaccard similarity. On the other hand, space-efficient computation of exact intersection sizes is particularly challenging in real-time. In this paper, we show how an efficient space-time trade-off can be achieved for exact set intersection counting, by combining state-of-the-art algorithms with precomputation and judicious use of compression. In addition, we show that the performance can be further improved by combining the best aspects of these algorithms. We present experimental evidence that realtime computation of exact intersection sizes is feasible with low memory overhead: we improve the mean query time of baseline approaches by over a factor of 100 using a data structure that takes merely twice the size of an inverted index. Overall, in our experiments, we achieve running times within the same order of magnitude as well-known approximation techniques.

Finding topical experts on micro-blogging sites, such as Twitter, is an essential information-seeking task. In this paper, we introduce an expert-finding algorithm for Twitter, which can be generalized to find topical experts in any social network with endorsement features.