The number of cleaned counter-examples across data sets and models is more than 30% of the total number of cleaned examples. FIM-based approaches outperform the LISSA estimator. FIM, which is difficult to store and invert. Figure 3 shows the results of the evaluation of Top Fisher, Practical Fisher and nearest neighbor (NN). As reported in the main text, Practical Fisher lags behind Top Fisher in all cases.

The number of cleaned counter-examples across data sets and models is more than 30% of the total number of cleaned examples. FIM-based approaches outperform the LISSA estimator. FIM, which is difficult to store and invert. Figure 3 shows the results of the evaluation of Top Fisher, Practical Fisher and nearest neighbor (NN). As reported in the main text, Practical Fisher lags behind Top Fisher in all cases.

Despite the growing promise of artificial intelligence (AI) in supporting decision-making across domains, fostering appropriate human reliance on AI remains a critical challenge. In this paper, we investigate the utility of exploring distance-based uncertainty scores for task delegation to AI and describe how these scores can be visualized through embedding representations for human-AI decision-making. After developing an AI-based system for physical stroke rehabilitation assessment, we conducted a study with 19 health professionals and 10 students in medicine/health to understand the effect of exploring distance-based uncertainty scores on users' reliance on AI. Our findings showed that distance-based uncertainty scores outperformed traditional probability-based uncertainty scores in identifying uncertain cases. In addition, after exploring confidence scores for task delegation and reviewing embedding-based visualizations of distance-based uncertainty scores, participants achieved an 8.20% higher rate of correct decisions, a 7.15% higher rate of changing their decisions to correct ones, and a 7.14% lower rate of incorrect changes after reviewing AI outputs than those reviewing probability-based uncertainty scores ($p<0.01$). Our findings highlight the potential of distance-based uncertainty scores to enhance decision accuracy and appropriate reliance on AI while discussing ongoing challenges for human-AI collaborative decision-making.

Recently, several methods have leveraged deep generative modeling to produce example-based explanations of decision algorithms for high-dimensional input data. Despite promising results, a disconnect exists between these methods and the classical explainability literature, which focuses on lower-dimensional data with semantically meaningful features. This conceptual and communication gap leads to misunderstandings and misalignments in goals and expectations. In this paper, we bridge this gap by proposing a novel probabilistic framework for local example-based explanations. Our framework integrates the critical characteristics of classical local explanation desiderata while being amenable to high-dimensional data and their modeling through deep generative models. Our aim is to facilitate communication, foster rigor and transparency, and improve the quality of peer discussion and research progress.

We tackle sequential learning under label noise in applications where a human supervisor can be queried to relabel suspicious examples. Existing approaches are flawed, in that they only relabel incoming examples that look "suspicious" to the model. As a consequence, those mislabeled examples that elude (or don't undergo) this cleaning step end up tainting the training data and the model with no further chance of being cleaned. We propose CINCER, a novel approach that cleans both new and past data by identifying \emph{pairs of mutually incompatible examples}. Whenever it detects a suspicious example, CINCER identifies a counter-example in the training set that - according to the model - is maximally incompatible with the suspicious example, and asks the annotator to relabel either or both examples, resolving this possible inconsistency.

A growing research explores the usage of AI explanations on user's decision phases for human-AI collaborative decision-making. However, previous studies found the issues of overreliance on `wrong' AI outputs. In this paper, we propose interactive example-based explanations to improve health professionals' onboarding with AI for their better reliance on AI during AI-assisted decision-making. We implemented an AI-based decision support system that utilizes a neural network to assess the quality of post-stroke survivors' exercises and interactive example-based explanations that systematically surface the nearest neighborhoods of a test/task sample from the training set of the AI model to assist users' onboarding with the AI model. To investigate the effect of interactive example-based explanations, we conducted a study with domain experts, health professionals to evaluate their performance and reliance on AI. Our interactive example-based explanations during onboarding assisted health professionals in having a better reliance on AI and making a higher ratio of making `right' decisions and a lower ratio of `wrong' decisions than providing only feature-based explanations during the decision-support phase. Our study discusses new challenges of assisting user's onboarding with AI for human-AI collaborative decision-making.

Tree-based machine learning models, such as decision trees and random forests, have been hugely successful in classification tasks primarily because of their predictive power in supervised learning tasks and ease of interpretation. Despite their popularity and power, these models have been found to produce unexpected or discriminatory outcomes. Given their overwhelming success for most tasks, it is of interest to identify sources of their unexpected and discriminatory behavior. However, there has not been much work on understanding and debugging tree-based classifiers in the context of fairness. We introduce FairDebugger, a system that utilizes recent advances in machine unlearning research to identify training data subsets responsible for instances of fairness violations in the outcomes of a random forest classifier. FairDebugger generates top-$k$ explanations (in the form of coherent training data subsets) for model unfairness. Toward this goal, FairDebugger first utilizes machine unlearning to estimate the change in the tree structures of the random forest when parts of the underlying training data are removed, and then leverages the Apriori algorithm from frequent itemset mining to reduce the subset search space. We empirically evaluate our approach on three real-world datasets, and demonstrate that the explanations generated by FairDebugger are consistent with insights from prior studies on these datasets.

A new approach to the local and global explanation is proposed. It is based on selecting a convex hull constructed for the finite number of points around an explained instance. The convex hull allows us to consider a dual representation of instances in the form of convex combinations of extreme points of a produced polytope. Instead of perturbing new instances in the Euclidean feature space, vectors of convex combination coefficients are uniformly generated from the unit simplex, and they form a new dual dataset. A dual linear surrogate model is trained on the dual dataset. The explanation feature importance values are computed by means of simple matrix calculations. The approach can be regarded as a modification of the well-known model LIME. The dual representation inherently allows us to get the example-based explanation. The neural additive model is also considered as a tool for implementing the example-based explanation approach. Many numerical experiments with real datasets are performed for studying the approach. The code of proposed algorithms is available.

A random forest prediction can be computed by the scalar product of the labels of the training examples and a set of weights that are determined by the leafs of the forest into which the test object falls; each prediction can hence be explained exactly by the set of training examples for which the weights are non-zero. The number of examples used in such explanations is shown to vary with the dimensionality of the training set and hyperparameters of the random forest algorithm. This means that the number of examples involved in each prediction can to some extent be controlled by varying these parameters. However, for settings that lead to a required predictive performance, the number of examples involved in each prediction may be unreasonably large, preventing the user to grasp the explanations. In order to provide more useful explanations, a modified prediction procedure is proposed, which includes only the top-weighted examples. An investigation on regression and classification tasks shows that the number of examples used in each explanation can be substantially reduced while maintaining, or even improving, predictive performance compared to the standard prediction procedure.

Can we preserve the accuracy of neural models while also providing faithful explanations? We present wrapper boxes, a general approach to generate faithful, example-based explanations for model predictions while maintaining predictive performance. After training a neural model as usual, its learned feature representation is input to a classic, interpretable model to perform the actual prediction. This simple strategy is surprisingly effective, with results largely comparable to those of the original neural model, as shown across three large pre-trained language models, two datasets of varying scale, four classic models, and four evaluation metrics. Moreover, because these classic models are interpretable by design, the subset of training examples that determine classic model predictions can be shown directly to users.