Machine learning models routinely automate decisions in applications like lending and hiring. In such settings, consumer protection rules require companies that deploy models to explain predictions to decision subjects. These rules are motivated, in part, by the belief that explanations can promote *recourse* by revealing information that individuals can use to contest or improve their outcomes. In practice, many companies comply with these rules by providing individuals with a list of the most important features for their prediction, which they identify based on feature importance scores from feature attribution methods such as SHAP or LIME. In this work, we show how these practices can undermine consumers by highlighting features that would not lead to an improved outcome and by explaining predictions that cannot be changed. We propose to address these issues by highlighting features based on their *responsiveness score*—i.e., the probability that an individual can attain a target prediction by changing a specific feature. We develop efficient methods to compute responsiveness scores for any model and any dataset. We conduct an extensive empirical study on the responsiveness of explanations in lending. Our results show that standard practices in consumer finance can backfire by presenting consumers with *reasons without recourse*, and demonstrate how our approach improves consumer protection by highlighting responsive features and identifying fixed predictions.

Machine learning models are routinely used to support decisions that affect individuals -- be it to screen a patient for a serious illness or to gauge their response to treatment. In these tasks, we are limited to learning models from datasets with noisy labels. In this paper, we study the instance-level impact of learning under label noise. We introduce a notion of regret for this regime, which measures the number of unforeseen mistakes due to noisy labels. We show that standard approaches to learning under label noise can return models that perform well at a population-level while subjecting individuals to a lottery of mistakes. We present a versatile approach to estimate the likelihood of mistakes at the individual-level from a noisy dataset by training models over plausible realizations of datasets without label noise. This is supported by a comprehensive empirical study of label noise in clinical prediction tasks. Our results reveal how failure to anticipate mistakes can compromise model reliability and adoption -- we demonstrate how we can address these challenges by anticipating and avoiding regretful decisions.

Many time series classification tasks, where labels vary over time, are affected by label noise that also varies over time. Such noise can cause label quality to improve, worsen, or periodically change over time. We first propose and formalize temporal label noise, an unstudied problem for sequential classification of time series. In this setting, multiple labels are recorded over time while being corrupted by a time-dependent noise function. We first demonstrate the importance of modeling the temporal nature of the label noise function and how existing methods will consistently underperform. We then propose methods to train noise-tolerant classifiers by estimating the temporal label noise function directly from data. We show that our methods lead to state-of-the-art performance under diverse types of temporal label noise on real-world datasets.

We introduce Concept Bottleneck Large Language Models (CB-LLMs), a novel framework for building inherently interpretable Large Language Models (LLMs). In contrast to traditional black-box LLMs that rely on limited post-hoc interpretations, CB-LLMs integrate intrinsic interpretability directly into the LLMs -- allowing accurate explanations with scalability and transparency. We build CB-LLMs for two essential NLP tasks: text classification and text generation. In text classification, CB-LLMs is competitive with, and at times outperforms, traditional black-box models while providing explicit and interpretable reasoning. For the more challenging task of text generation, interpretable neurons in CB-LLMs enable precise concept detection, controlled generation, and safer outputs. The embedded interpretability empowers users to transparently identify harmful content, steer model behavior, and unlearn undesired concepts -- significantly enhancing the safety, reliability, and trustworthiness of LLMs, which are critical capabilities notably absent in existing language models.

Many applications in machine learning and decision-making rely on procedures to aggregate human preferences. In such tasks, individual express ordinal preferences over a set of items through votes, ratings, or pairwise comparisons. We then summarize their collective preferences as a ranking. Standard methods for preference aggregation are designed to return rankings that arbitrate individual disagreements in ways that are faithful and fair. In this work, we introduce a paradigm for *selective aggregation*, where we can avoid the need to arbitrate dissent by abstaining from comparison. We summarize collective preferences as a *selective ranking* -- i.e., a partial order where we can only compare items where at least $100\cdot(1 - \tau)\%$ of individuals agree. We develop algorithms to build selective rankings that achieve all possible trade-offs between comparability and disagreement, and derive formal guarantees on their safety and stability. We conduct an extensive set of experiments on real-world datasets to benchmark our approach and demonstrate its functionality. Our results show selective aggregation can promote transparency and robustness by revealing disagreement and abstaining from arbitration.

Machine learning models can assign fixed predictions that preclude individuals from changing their outcome. Existing approaches to audit fixed predictions do so on a pointwise basis, which requires access to an existing dataset of individuals and may fail to anticipate fixed predictions in out-of-sample data. This work presents a new paradigm to identify fixed predictions by finding confined regions of the feature space in which all individuals receive fixed predictions. This paradigm enables the certification of recourse for out-of-sample data, works in settings without representative datasets, and provides interpretable descriptions of individuals with fixed predictions. We develop a fast method to discover confined regions for linear classifiers using mixed-integer quadratically constrained programming. We conduct a comprehensive empirical study of confined regions across diverse applications. Our results highlight that existing pointwise verification methods fail to anticipate future individuals with fixed predictions, while our method both identifies them and provides an interpretable description.

Machine learning models now automate decisions in applications where we may wish to provide recourse to adversely affected individuals. In practice, existing methods to provide recourse return actions that fail to account for latent characteristics that are not captured in the model (e.g., age, sex, marital status). In this paper, we study how the cost and feasibility of recourse can change across these latent groups. We introduce a notion of group-level plausibility to identify groups of individuals with a shared set of latent characteristics. We develop a general-purpose clustering procedure to identify groups from samples. Further, we propose a constrained optimization approach to learn models that equalize the cost of recourse over latent groups. We evaluate our approach through an empirical study on simulated and real-world datasets, showing that it can produce models that have better performance in terms of overall costs and feasibility at a group level.

We propose a new approach to promote safety in classification tasks with concept annotations. Our approach – called a *conceptual safeguard* – acts as a verification layer for models that predict a target outcome by first predicting the presence of intermediate concepts. Given this architecture, a safeguard ensures that a model meets a minimal level of accuracy by abstaining from uncertain predictions. In contrast to a standard selective classifier, a safeguard provides an avenue to improve coverage by allowing a human to confirm the presence of uncertain concepts on instances on which it abstains. We develop methods to build safeguards that maximize coverage without compromising safety, namely techniques to propagate the uncertainty in concept predictions and to flag salient concepts for human review. We benchmark our approach on a collection of real-world and synthetic datasets, showing that it can improve performance and coverage in deep learning tasks.

Machine learning models are often used to decide who receives a loan, a job interview, or a public benefit. Models in such settings use features without considering their *actionability*. As a result, they can assign predictions that are \emph{fixed} -- meaning that individuals who are denied loans and interviews are, in fact, *precluded from access* to credit and employment. In this work, we introduce a procedure called *recourse verification* to test if a model assigns fixed predictions to its decision subjects. We propose a model-agnostic approach for verification with *reachable sets* -- i.e., the set of all points that a person can reach through their actions in feature space. We develop methods to construct reachable sets for discrete feature spaces, which can certify the responsiveness of *any model* by simply querying its predictions. We conduct a comprehensive empirical study on the infeasibility of recourse on datasets from consumer finance. Our results highlight how models can inadvertently preclude access by assigning fixed predictions and underscore the need to account for actionability in model development.

Modern recommender systems may output considerably different recommendations due to small perturbations in the training data. Changes in the data from a single user will alter the recommendations as well as the recommendations of other users. In applications like healthcare, housing, and finance, this sensitivity can have adverse effects on user experience. We propose a method to stabilize a given recommender system against such perturbations. This is a challenging task due to (1) the lack of a “reference” rank list that can be used to anchor the outputs; and (2) the computational challenges in ensuring the stability of rank lists with respect to all possible perturbations of training data. Our method, FINEST , overcomes these challenges by obtaining reference rank lists from a given recommendation model and then fine-tuning the model under simulated perturbation scenarios with rank-preserving regularization on sampled items. Our experiments on real-world datasets demonstrate that FINEST can ensure that recommender models output stable recommendations under a wide range of different perturbations without compromising next-item prediction accuracy.

Machine learning models are often used to inform real world risk assessment tasks: predicting consumer default risk, predicting whether a person suffers from a serious illness, or predicting a person's risk to appear in court. Given multiple models that perform almost equally well for a prediction task, to what extent do predictions vary across these models? If predictions are relatively consistent for similar models, then the standard approach of choosing the model that optimizes a penalized loss suffices. But what if predictions vary significantly for similar models? In machine learning, this is referred to as predictive multiplicity i.e. the prevalence of conflicting predictions assigned by near-optimal competing models. In this paper, we present a framework for measuring predictive multiplicity in probabilistic classification (predicting the probability of a positive outcome). We introduce measures that capture the variation in risk estimates over the set of competing models, and develop optimization-based methods to compute these measures efficiently and reliably for convex empirical risk minimization problems. We demonstrate the incidence and prevalence of predictive multiplicity in real-world tasks. Further, we provide insight into how predictive multiplicity arises by analyzing the relationship between predictive multiplicity and data set characteristics (outliers, separability, and majority-minority structure). Our results emphasize the need to report predictive multiplicity more widely.

Machine learning models are often personalized with categorical attributes that define groups. In this work, we show that personalization with *group attributes* can inadvertently reduce performance at a *group level* -- i.e., groups may receive unnecessarily inaccurate predictions by sharing their personal characteristics. We present formal conditions to ensure the *fair use* of group attributes in a prediction task, and describe how they can be checked by training one additional model. We characterize how fair use conditions be violated due to standard practices in model development, and study the prevalence of fair use violations in clinical prediction tasks. Our results show that personalization often fails to produce a tailored performance gain for every group who reports personal data, and underscore the need to evaluate fair use when personalizing models with characteristics that are protected, sensitive, self-reported, or costly to acquire.

Machine learning models are often personalized based on information that is protected, sensitive, self-reported, or costly to acquire. These models use information about people, but do not facilitate nor inform their *consent*. Individuals cannot opt out of reporting information that a model needs to personalize their predictions nor tell if they benefit from personalization in the first place. We introduce a new family of prediction models, called participatory systems, that let individuals opt into personalization at prediction time. We present a model-agnostic algorithm to learn participatory systems for supervised learning tasks where models are personalized with categorical group attributes. We conduct a comprehensive empirical study of participatory systems in clinical prediction tasks, comparing them to common approaches for personalization and imputation. Our results show that participatory systems can facilitate and inform consent in a way that improves performance and privacy across all groups who report personal data.

Prediction models can exhibit sensitivity with respect to training data: small changes in the training data can produce models that assign conflicting predictions to individual data points during test time. In this work, we study this sensitivity in recommender systems, where users' recommendations are drastically altered by minor perturbations in other unrelated users' interactions. We introduce a measure of stability for recommender systems, called Rank List Sensitivity (RLS), which measures how rank lists generated by a given recommender system at test time change as a result of a perturbation in the training data. We develop a method, CASPER, which uses cascading effect to identify the minimal and systematical perturbation to induce higher instability in a recommender system. Experiments on four datasets show that recommender models are overly sensitive to minor perturbations introduced randomly or via CASPER - even perturbing one random interaction of one user drastically changes the recommendation lists of all users.Importantly, with CASPER perturbation, the models generate more unstable recommend ations for low-accuracy users (i.e., those who receive low-quality recommendations) than high-accuracy ones.

Machine learning models are often personalized by using group attributes that encode personal characteristics (e.g., sex, age group, HIV status). In such settings, individuals expect to receive more accurate predictions in return for disclosing group attributes to the personalized model. We study when we can tell that a personalized model upholds this principle for every group who provides personal data. We introduce a metric called the benefit of personalization (BoP) to measure the smallest gain in accuracy that any group expects to receive from a personalized model. We describe how the BoP can be used to carry out basic routines to audit a personalized model, including: (i) hypothesis tests to check that a personalized model improves performance for every group; (ii) estimation procedures to bound the minimum gain in personalization. We characterize the reliability of these routines in a finite-sample regime and present minimax bounds on both the probability of error for BoP hypothesis tests and the mean-squared error of BoP estimates. Our results show that we can only claim that personalization improves performance for each group who provides data when we explicitly limit the number of group attributes used by a personalized model. In particular, we show that it is impossible to reliably verify that a personalized classifier with $k \geq 19$ binary group attributes will benefit every group who provides personal data using a dataset of $n = 8\times10^9$ samples -- one for each person in the world.

Checklists are simple decision aids that are often used to promote safety and reliability in clinical applications. In this paper, we present a method to learn checklists for clinical decision support. We represent predictive checklists as discrete linear classifiers with binary features and unit weights. We then learn globally optimal predictive checklists from data by solving an integer programming problem. Our method allows users to customize checklists to obey complex constraints, including constraints to enforce group fairness and to binarize real-valued features at training time. In addition, it pairs models with an optimality gap that can inform model development and determine the feasibility of learning sufficiently accurate checklists on a given dataset. We pair our method with specialized techniques that speed up its ability to train a predictive checklist that performs well and has a small optimality gap. We benchmark the performance of our method on seven clinical classification problems, and demonstrate its practical benefits by training a short-form checklist for PTSD screening. Our results show that our method can fit simple predictive checklists that perform well and that can easily be customized to obey a rich class of custom constraints.

Prediction problems often admit competing models that perform almost equally well. This effect challenges key assumptions in machine learning when competing models assign conflicting predictions. In this paper, we define predictive multiplicity as the ability of a prediction problem to admit competing models with conflicting predictions. We introduce measures to evaluate the severity of predictive multiplicity, and develop integer programming tools to compute these measures exactly for linear classification problems. We apply our tools to measure predictive multiplicity in recidivism prediction problems. Our results show that real-world datasets may admit competing models that assign wildly conflicting predictions, and motivate the need to report predictive multiplicity in model development.

In domains such as medicine, it can be acceptable for machine learning models to include sensitive attributes such as gender and ethnicity. In this work, we argue that when there is this kind of treatment disparity, then it should be in the best interest of each group. Drawing on ethical principles such as beneficence ("do the best") and non-maleficence ("do no harm"), we show how to use sensitive attributes to train decoupled classifiers that satisfy preference guarantees. These guarantees ensure the majority of individuals in each group prefer their assigned classifier to (i) a pooled model that ignores group membership (rationality), and (ii) the model assigned to any other group (envy-freeness). We introduce a recursive procedure that adaptively selects group attributes for decoupling, and present formal conditions to ensure preference guarantees in terms of generalization error. We validate the effectiveness of the procedure on real-world datasets, showing that it improves accuracy without violating preference guarantees on test data.

When the performance of a machine learning model varies over groups defined by sensitive attributes (e.g., gender or ethnicity), the performance disparity can be expressed in terms of the probability distributions of the input and output variables over each group. In this paper, we exploit this fact to reduce the disparate impact of a fixed classification model over a population of interest. Given a black-box classifier, we aim to eliminate the performance gap by perturbing the distribution of input variables for the disadvantaged group. We refer to the perturbed distribution as a counterfactual distribution, and characterize its properties for common fairness criteria. We introduce a descent algorithm to learn a counterfactual distribution from data. We then discuss how the estimated distribution can be used to build a data preprocessor that can reduce disparate impact without training a new model. We validate our approach through experiments on real-world datasets, showing that it can repair different forms of disparity without a significant drop in accuracy.

Risk scores are simple classification models that let users make quick risk predictions by adding and subtracting a few small numbers. These models are widely used in medicine and criminal justice, but are difficult to learn from data because they need to be calibrated, sparse, use small integer coefficients, and obey application-specific constraints. In this paper, we introduce a machine learning method to learn risk scores. We formulate the risk score problem as a mixed integer nonlinear program, and present a cutting plane algorithm to recover its optimal solution. We improve our algorithm with specialized techniques that generate feasible solutions, narrow the optimality gap, and reduce data-related computation. Our algorithm can train risk scores in a way that scales linearly in the number of samples in a dataset, and that allows practitioners to address application-specific constraints without parameter tuning or post-processing. We benchmark the performance of different methods to learn risk scores on publicly available datasets, comparing risk scores produced by our method to risk scores built using methods that are used in practice. We also discuss the practical benefits of our method through a real-world application where we build a customized risk score for ICU seizure prediction in collaboration with the Massachusetts General Hospital.

In the context of machine learning, disparate impact refers to a form of systematic discrimination whereby the output distribution of a model depends on the value of a sensitive attribute (e.g., race or gender). In this paper, we propose an information-theoretic framework to analyze the disparate impact of a binary classification model. We view the model as a fixed channel, and quantify disparate impact as the divergence in output distributions over two groups. Our aim is to find a correction function that can perturb the input distributions of each group to align their output distributions. We present an optimization problem that can be solved to obtain a correction function that will make the output distributions statistically indistinguishable. We derive closed-form expressions to efficiently compute the correction function, and demonstrate the benefits of our framework on a recidivism prediction problem based on the ProPublica COMPAS dataset.

Risk scores are simple classification models that let users quickly assess risk by adding, subtracting, and multiplying a few small numbers. Such models are widely used in healthcare and criminal justice, but are often built ad hoc. In this paper, we present a principled approach to learn risk scores that are fully optimized for feature selection, integer coefficients, and operational constraints. We formulate the risk score problem as a mixed integer nonlinear program, and present a new cutting plane algorithm to efficiently recover its optimal solution. Our approach can fit optimized risk scores in a way that scales linearly with the sample size of a dataset, provides a proof of optimality, and obeys complex constraints without parameter tuning. We illustrate these benefits through an extensive set of numerical experiments, and an application where we build a customized risk score for ICU seizure prediction.

Stochastic programming models are large-scale optimization problems that are used to facilitate decision making under uncertainty. Optimization algorithms for such problems need to evaluate the expected future costs of current decisions, often referred to as the recourse function. In practice, this calculation is computationally difficult as it requires the evaluation of a multidimensional integral whose integrand is an optimization problem. In turn, the recourse function has to be estimated using techniques such as scenario trees or Monte Carlo methods, both of which require numerous functional evaluations to produce accurate results for large-scale problems with multiple periods and high-dimensional uncertainty. In this work, we introduce an importance sampling framework for stochastic programming that can produce accurate estimates of the recourse function using a small number of samples. Our framework combines Markov chain Monte Carlo methods with kernel density estimation algorithms to build a nonparametric importance sampling distribution, which can then be used to produce a lower-variance estimate of the recourse function. We demonstrate the increased accuracy and efficiency of our approach using variants of well-known multistage stochastic programming problems. Our numerical results show that our framework produces more accurate estimates of the optimal value of stochastic programming models, especially for problems with moderate variance, multimodal, or rare-event distributions.