Explanations are an important tool for gaining insights into model behavior, calibrating user trust, and ensuring compliance.The past few years have seen a flurry of methods for generating explanations, many of which involve computing model gradients or solving specially designed optimization problems.Owing to the remarkable reasoning abilities of LLMs, selfexplanation, i.e., prompting the model to explain its outputs, has recently emerged as a new paradigm.We study a specific type of self-explanation, self-generated counterfactual explanations (SCEs).We test LLMs' ability to generate SCEs across families, sizes, temperatures, and datasets.We find that LLMs sometimes struggle to generate SCEs.When they do, their prediction often does not agree with their own counterfactual reasoning. github.com/aisoc-lab/llm-sces

Elisabeth Kirsten, Ivan Habernal, Vedant Nanda, Muhammad Bilal Zafar. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Static code analyzers are now a common part of the codereview process. These automated tools integrate into the code review process by commenting on code changes and suggesting improvements, in the same way as human reviewers. The comments made by static analyzers often trigger a conversation between developers to align on if and how the issue should be fixed. Because developers rarely give feedback directly to the tool, understanding the sentiment and intent in the conversation triggered by the tool comments can be used to measure the usefulness of the static analyzer.

While large language models (LLMs) have taken great strides towards helping humans with a plethora of tasks, hallucinations remain a major impediment towards gaining user trust. The fluency and coherence of model generations even when hallucinating makes detection a difficult task. In this work, we explore if the artifacts associated with the model generations can provide hints that the generation will contain hallucinations. Specifically, we probe LLMs at 1) the inputs via Integrated Gradients based token attribution, 2) the outputs via the Softmax probabilities, and 3) the internal state via self-attention and fully-connected layer activations for signs of hallucinations on open-ended question answering tasks. Our results show that the distributions of these artifacts tend to differ between hallucinated and non-hallucinated generations. Building on this insight, we train binary classifiers that use these artifacts as input features to classify model generations into hallucinations and non-hallucinations. These hallucination classifiers achieve up to 0.80 AUROC. We also show that tokens preceding a hallucination can already predict the subsequent hallucination even before it occurs.

We revisit the problem of fair principal component analysis (PCA), where the goal is to learn the best low-rank linear approximation of the data that obfuscates demographic information. We propose a conceptually simple approach that allows for an analytic solution similar to standard PCA and can be kernelized. Our methods have the same complexity as standard PCA, or kernel PCA, and run much faster than existing methods for fair PCA based on semidefinite programming or manifold optimization, while achieving similar results.

While using vast amounts of training data and sophisticated models has enhanced the predictive performance of Machine Learning (ML) and Artificial Intelligence (AI) solutions, it has also led to an increased difficulty in comprehending their predictions. The ability to explain predictions is often one of the primary desiderata for adopting AI and ML solutions [6, 13]. The desire for explainability has led to a rapidly growing body of literature on explainable AI (XAI) and has also resulted in the development of hundreds of XAI methods targeting different domains (e.g., finance, healthcare), applications (e.g., model debugging, actionable recourse), data modalities (e.g., tabular data, images), models (e.g., transformers, convolutional neural networks) and stakeholders (e.g., end-users, regulatory authorities, data scientists). The goal of this tutorial is to present a comprehensive overview of the XAI field to the participants. As a hands-on tutorial, we will showcase state-of-the-art methods that can be used for different data modalities and contexts to extract the right abstractions for interpretation. We will also cover common pitfalls when using explanations, e.g., misrepresentation, and lack of robustness of explanations.

We initiate the study of fairness for ordinal regression. We adapt two fairness notions previously considered in fair ranking and propose a strategy for training a predictor that is approximately fair according to either notion. Our predictor has the form of a threshold model, composed of a scoring function and a set of thresholds, and our strategy is based on a reduction to fair binary classification for learning the scoring function and local search for choosing the thresholds. We provide generalization guarantees on the error and fairness violation of our predictor, and we illustrate the effectiveness of our approach in extensive experiments.

Deep neural networks (DNNs) are known to be highly vulnerable to adversarial examples (AEs) that include malicious perturbations. Assumptions about the statistical differences between natural and adversarial inputs are commonplace in many detection techniques. As a best practice, AE detectors are evaluated against ’adaptive’ attackers who actively perturb their inputs to avoid detection. Due to the difficulties in designing adaptive attacks, however, recent work suggests that most detectors have incomplete evaluation. We aim to fill this gap by designing a generic adaptive attack against detectors: the ’statistical indistinguishability attack’ (SIA). SIA optimizes a novel objective to craft adversarial examples (AEs) that follow the same distribution as the natural inputs with respect to DNN representations. Our objective targets all DNN layers simultaneously as we show that AEs being indistinguishable at one layer might fail to be so at other layers. SIA is formulated around evading distributional detectors that inspect a set of AEs as a whole and is also effective against four individual AE detectors, two dataset shift detectors, and an out-of-distribution sample detector, curated from published works. This suggests that SIA can be a reliable tool for evaluating the security of a range of detectors.

With the increasing adoption of machine learning (ML) models and systems in high-stakes settings across different industries, guaranteeing a model's performance after deployment has become crucial. Monitoring models in production is a critical aspect of ensuring their continued performance and reliability. We present Amazon SageMaker Model Monitor, a fully managed service that continuously monitors the quality of machine learning models hosted on Amazon SageMaker. Our system automatically detects data, concept, bias, and feature attribution drift in models in real-time and provides alerts so that model owners can take corrective actions and thereby maintain high quality models. We describe the key requirements obtained from customers, system design and architecture, and methodology for detecting different types of drift. Further, we provide quantitative evaluations followed by use cases, insights, and lessons learned from more than two years of production deployment.

Given the increasing importance of machine learning (ML) in our lives, several algorithmic fairness techniques have been proposed to mitigate biases in the outcomes of the ML models. However, most of these techniques are specialized to cater to a single family of ML models and a specific definition of fairness, limiting their adaptibility in practice. We introduce a general constrained Bayesian optimization (BO) framework to optimize the performance of any ML model while enforcing one or multiple fairness constraints. BO is a model-agnostic optimization method that has been successfully applied to automatically tune the hyperparameters of ML models. We apply BO with fairness constraints to a range of popular models, including random forests, gradient boosting, and neural networks, showing that we can obtain accurate and fair solutions by acting solely on the hyperparameters. We also show empirically that our approach is competitive with specialized techniques that enforce model-specific fairness constraints, and outperforms preprocessing methods that learn fair representations of the input data. Moreover, our method can be used in synergy with such specialized fairness techniques to tune their hyperparameters. Finally, we study the relationship between fairness and the hyperparameters selected by BO. We observe a correlation between regularization and unbiased models, explaining why acting on the hyperparameters leads to ML models that generalize well and are fair.

Understanding the predictions made by machine learning (ML) models and their potential biases remains a challenging and labor-intensive task that depends on the application, the dataset, and the specific model. We present Amazon SageMaker Clarify, an explainability feature for Amazon SageMaker that launched in December 2020, providing insights into data and ML models by identifying biases and explaining predictions. It is deeply integrated into Amazon SageMaker, a fully managed service that enables data scientists and developers to build, train, and deploy ML models at any scale. Clarify supports bias detection and feature importance computation across the ML lifecycle, during data preparation, model evaluation, and post-deployment monitoring. We outline the desiderata derived from customer input, the modular architecture, and the methodology for bias and explanation computations. Further, we describe the technical challenges encountered and the tradeoffs we had to make. For illustration, we discuss two customer use cases. We present our deployment results including qualitative customer feedback and a quantitative evaluation. Finally, we summarize lessons learned, and discuss best practices for the successful adoption of fairness and explanation tools in practice.

The use of algorithmic (learning-based) decision making in scenarios that affect human lives has motivated a number of recent studies to investigate such decision making systems for potential unfairness, such as discrimination against subjects based on their sensitive features like gender or race. However, when judging the fairness of a newly designed decision making system, these studies have overlooked an important influence on people's perceptions of fairness, which is how the new algorithm changes the status quo, i.e., decisions of the existing decision making system. Motivated by extensive literature in behavioral economics and behavioral psychology (prospect theory), we propose a notion of fair updates that we refer to as loss-averse updates. Loss-averse updates constrain the updates to yield improved (more beneficial) outcomes to subjects compared to the status quo. We propose tractable proxy measures that would allow this notion to be incorporated in the training of a variety of linear and non-linear classifiers. We show how our proxy measures can be combined with existing measures for training nondiscriminatory classifiers.Our evaluation using synthetic and real-world datasets demonstrates that the proposed proxy measures are effective for their desired tasks.

Users frequently use search systems on the Web as well as online social media to learn about ongoing events and public opinion on personalities. Prior studies have shown that the top-ranked results returned by these search engines can shape user opinion about the topic (e.g., event or person) being searched. In case of polarizing topics like politics, where multiple competing perspectives exist, the political bias in the top search results can play a significant role in shaping public opinion towards (or away from) certain perspectives. Given the considerable impact that search bias can have on the user, we propose a generalizable search bias quantification framework that not only measures the political bias in ranked list output by the search system but also decouples the bias introduced by the different sources—input data and ranking system. We apply our framework to study the political bias in searches related to 2016 US Presidential primaries in Twitter social media search and find that both input data and ranking system matter in determining the final search output bias seen by the users. And finally, we use the framework to compare the relative bias for two popular search systems—Twitter social media search and Google web search—for queries related to politicians and political events. We end by discussing some potential solutions to signal the bias in the search results to make the users more aware of them.

Algorithmic decision making is employed in an increasing number of real-world applicationstions to aid human decision making. While it has shown considerable promise in terms of improved decision accuracy, in some scenarios, its outcomes have been also shown to impose an unfair (dis)advantage on people from certain social groups (e.g., women, blacks). In this context, there is a need for computational techniques to limit unfairness in algorithmic decision making. In this work, we take a step forward to fulfill that need and introduce a flexible constraint-based framework to enable the design of fair margin-based classifiers. The main technical innovation of our framework is a general and intuitive measure of decision boundary unfairness, which serves as a tractable proxy to several of the most popular computational definitions of unfairness from the literature. Leveraging our measure, we can reduce the design of fair margin-based classifiers to adding tractable constraints on their decision boundaries. Experiments on multiple synthetic and real-world datasets show that our framework is able to successfully limit unfairness, often at a small cost in terms of accuracy.

With widespread use of machine learning methods in numerous domains involving humans, several studies have raised questions about the potential for unfairness towards certain individuals or groups. A number of recent works have proposed methods to measure and eliminate unfairness from machine learning models. However, most of this work has focused on only one dimension of fair decision making: distributive fairness, i.e., the fairness of the decision outcomes. In this work, we leverage the rich literature on organizational justice and focus on another dimension of fair decision making: procedural fairness, i.e., the fairness of the decision making process. We propose measures for procedural fairness that consider the input features used in the decision process, and evaluate the moral judgments of humans regarding the use of these features. We operationalize these measures on two real world datasets using human surveys on the Amazon Mechanical Turk (AMT) platform, demonstrating that our measures capture important properties of procedurally fair decision making. We provide fast submodular mechanisms to optimize the tradeoff between procedural fairness and prediction accuracy. On our datasets, we observe empirically that procedural fairness may be achieved with little cost to outcome fairness, but that some loss of accuracy is unavoidable.

Discrimination via algorithmic decision making has received considerable attention. Prior work largely focuses on defining conditions for fairness, but does not define satisfactory measures of algorithmic unfairness. In this paper, we focus on the following question: Given two unfair algorithms, how should we determine which of the two is more unfair? Our core idea is to use existing inequality indices from economics to measure how unequally the outcomes of an algorithm benefit different individuals or groups in a population. Our work offers a justified and general framework to compare and contrast the (un)fairness of algorithmic predictors. This unifying approach enables us to quantify unfairness both at the individual and the group level. Further, our work reveals overlooked tradeoffs between different fairness notions: using our proposed measures, the overall individual-level unfairness of an algorithm can be decomposed into a between-group and a within-group component. Earlier methods are typically designed to tackle only between-group un- fairness, which may be justified for legal or other reasons. However, we demonstrate that minimizing exclusively the between-group component may, in fact, increase the within-group, and hence the overall unfairness. We characterize and illustrate the tradeoffs between our measures of (un)fairness and the prediction accuracy.

Algorithmic decision making systems are ubiquitous across a wide variety of online as well as offline services. These systems rely on complex learning methods and vast amounts of data to optimize the service functionality, satisfaction of the end user and profitability. However, there is a growing concern that these automated decisions can lead, even in the absence of intent, to a lack of fairness, i.e., their outcomes can disproportionately hurt (or, benefit) particular groups of people sharing one or more sensitive attributes (e.g., race, sex). In this paper, we introduce a flexible mechanism to design fair classifiers by leveraging a novel intuitive measure of decision boundary (un)fairness. We instantiate this mechanism with two well-known classifiers, logistic regression and support vector machines, and show on real-world data that our mechanism allows for a fine-grained control on the degree of fairness, often at a small cost in terms of accuracy.

Search systems in online social media sites are frequently used to find information about ongoing events and people. For topics with multiple competing perspectives, such as political events or political candidates, bias in the top ranked results significantly shapes public opinion. However, bias does not emerge from an algorithm alone. It is important to distinguish between the bias that arises from the data that serves as the input to the ranking system and the bias that arises from the ranking system itself. In this paper, we propose a framework to quantify these distinct biases and apply this framework to politics-related queries on Twitter. We found that both the input data and the ranking system contribute significantly to produce varying amounts of bias in the search results and in different ways. We discuss the consequences of these biases and possible mechanisms to signal this bias in social media search systems' interfaces.

The adoption of automated, data-driven decision making in an ever expanding range of applications has raised concerns about its potential unfairness towards certain social groups. In this context, a number of recent studies have focused on defining, detecting, and removing unfairness from data-driven decision systems. However, the existing notions of fairness, based on parity (equality) in treatment or outcomes for different social groups, tend to be quite stringent, limiting the overall decision making accuracy. In this paper, we draw inspiration from the fair-division and envy-freeness literature in economics and game theory and propose preference-based notions of fairness -- given the choice between various sets of decision treatments or outcomes, any group of users would collectively prefer its treatment or outcomes, regardless of the (dis)parity as compared to the other groups. Then, we introduce tractable proxies to design margin-based classifiers that satisfy these preference-based notions of fairness. Finally, we experiment with a variety of synthetic and real-world datasets and show that preference-based fairness allows for greater decision accuracy than parity-based fairness.

Automated data-driven decision making systems are increasingly being used to assist, or even replace humans in many settings. These systems function by learning from historical decisions, often taken by humans. In order to maximize the utility of these systems (or, classifiers), their training involves minimizing the errors (or, misclassifications) over the given historical data. However, it is quite possible that the optimally trained classifier makes decisions for people belonging to different social groups with different misclassification rates (e.g., misclassification rates for females are higher than for males), thereby placing these groups at an unfair disadvantage. To account for and avoid such unfairness, in this paper, we introduce a new notion of unfairness, disparate mistreatment, which is defined in terms of misclassification rates. We then propose intuitive measures of disparate mistreatment for decision boundary-based classifiers, which can be easily incorporated into their formulation as convex-concave constraints. Experiments on synthetic as well as real world datasets show that our methodology is effective at avoiding disparate mistreatment, often at a small cost in terms of accuracy.

Extracting news on specific topics from the Twitter microblogging site poses formidable challenges, which include handling millions of tweets posted daily, judging topicality and importance of tweets, and ensuring trustworthiness of results in the face of spam. To date, all scalable approaches have relied on crowd wisdom, i.e., keyword-matching on the global tweet stream to gather relevant tweets, and crowd- endorsements to judge the importance of tweets. We propose a fundamentally different methodology -- for a given topic, we identify trustworthy experts on the topic, and extract news-stories that are most popular among the experts. Comparing the crowd-based and expert-based methodologies, we demonstrate that the news-stories obtained by our methodology (i) have higher relevance for a wide variety of topics, (ii) achieve very high coverage of important news-stories posted globally in Twitter, and (iii) are far more trustworthy. Using our methodology, we implemented and publicly deployed a topical news system for Twitter, which can extract news-stories on thousands of topics.

Abstract The decentralized I owe you (IOU) transaction network Ripple is gaining prominence as a fast, low-cost and efficient method for performing same and cross-currency payments. Ripple keeps track of IOU credit its users have granted to their business partners or friends, and settles transactions between two connected Ripple wallets by appropriately changing credit values on the connecting paths. Similar to cryptocurrencies such as Bitcoin, while the ownership of the wallets is implicitly pseudonymous in Ripple, IOU credit links and transaction flows between wallets are publicly available in an online ledger. In this paper, we present the first thorough study that analyzes this globally visible log and characterizes the privacy issues with the current Ripple network. In particular, we define two novel heuristics and perform heuristic clustering to group wallets based on observations on the Ripple network graph. We then propose reidentification mechanisms to deanonymize the operators of those clusters and show how to reconstruct the financial activities of deanonymized Ripple wallets. Our analysis motivates the need for better privacy-preserving payment mechanisms for Ripple and characterizes the privacy challenges faced by the emerging credit networks.

Analysis of content streams gathered from social networking sites such as Twitter has several applications ranging from content search and recommendation, news detection to business analytics. However, processing large amounts of data generated on these sites in real-time poses a difficult challenge. To cope with the data deluge, analytics companies and researchers are increasingly resorting to sampling. In this article, we investigate the crucial question of how to sample content streams generated by users in online social networks . The traditional method is to randomly sample all the data. For example, most studies using Twitter data today rely on the 1% and 10% randomly sampled streams of tweets that are provided by Twitter. In this paper, we analyze a different sampling methodology, one where content is gathered only from a relatively small sample (<1%) of the user population, namely, the expert users . Over the duration of a month, we gathered tweets from over 500,000 Twitter users who are identified as experts on a diverse set of topics, and compared the resulting expert sampled tweets with the 1% randomly sampled tweets provided publicly by Twitter. We compared the sampled datasets along several dimensions, including the popularity, topical diversity, trustworthiness, and timeliness of the information contained within them, and on the sentiment/opinion expressed on specific topics. Our analysis reveals several important differences in data obtained through the different sampling methodologies, which have serious implications for applications such as topical search, trustworthy content recommendations, breaking news detection, and opinion mining.

We present a semantic methodology to identify topical groups in Twitter on a large number of topics, each consisting of users who are experts on or interested in a specific topic. Early studies investigating the nature of Twitter suggest that it is a social media platform consisting of a relatively small section of elite users, producing information on a few popular topics such as media, politics, and music, and the general population consuming it. We show that this characterization ignores a rich set of highly specialized topics, ranging from geology, neurology, to astrophysics and karate - each being discussed by their own topical groups. We present a detailed characterization of these topical groups based on their network structures and tweeting behaviors. Analyzing these groups on the backdrop of the common identity and bond theory in social sciences shows that these groups exhibit characteristics of topical-identity based groups, rather than social-bond based ones.

We propose a novel mechanism to infer topics of interest of individual users in the Twitter social network. We observe that in Twitter, a user generally follows experts on various topics of her interest in order to acquire information on those topics. We use a methodology based on social annotations (proposed earlier by us) to first deduce the topical expertise of popular Twitter users, and then transitively infer the interests of the users who follow them. This methodology is a sharp departure from the traditional techniques of inferring interests of a user from the tweets that she posts or receives. We show that the topics of interest inferred by the proposed methodology are far superior than the topics extracted by state-of-the-art techniques such as using topic models (Labeled LDA) on tweets. Based upon the proposed methodology, we build a system Who Likes What, which can infer the interests of millions of Twitter users. To our knowledge, this is the first system that can infer interests for Twitter users at such scale. Hence, this system would be particularly beneficial in developing personalized recommender services over the Twitter platform.

Several applications today rely upon content streams crowd-sourced from online social networks. Since real-time processing of large amounts of data generated on these sites is difficult, analytics companies and researchers are increasingly resorting to sampling. In this paper, we investigate the crucial question of how to sample the data generated by users in social networks. The traditional method is to randomly sample all the data. We analyze a different sampling methodology, where content is gathered only from a relatively small subset (< 1%) of the user population namely, the expert users. Over the duration of a month, we gathered tweets from over 500,000 Twitter users who are identified as experts on a diverse set of topics, and compared the resulting expert-sampled tweets with the 1% randomly sampled tweets provided publicly by Twitter. We compared the sampled datasets along several dimensions, including the diversity, timeliness, and trustworthiness of the information contained within them, and find important differences between the datasets. Our observations have major implications for applications such as topical search, trustworthy content recommendations, and breaking news detection.

Today router buffers are sized according to the well-known Bandwidth-Delay Product (BDP) rule, which uses the average round-trip time (RTT) of flows traversing a router. The BDP rule not only leads to large queueing delays, but also imposes “one (buffer) size fits all” philosophy for flows exhibiting a large variation in RTT. When short and long RTT flows compete at a single buffer, they may adversely affect each other in throughput and delay. We propose SplitBuff, a scheme using which short RTT flows achieve low delay and long RTT flows achieve high throughput, without requiring any protocol modifications. With SplitBuff, a router splits a buffer into multiple buffers of varying sizes and maps flows onto these buffers based on their RTTs. We describe SplitBuff and evaluate its performance using extensive ns-2 simulations to demonstrate its effectiveness.