Coverage-based prime implicant explanations are formal explanations offering a number of valuable assets, especially in terms of faithfulness and generality. Unfortunately, deriving a coverage-based prime implicant explanation for an instance is computationally hard in the general case (the problem of identifying such an explanation being at the second level of the polynomial hierarchy). In this paper, we focus on the computation of a coverage-based prime implicant explanation for an instance given a tree-based model. We show that the specific nature of the domain theory linking the Boolean conditions in such models makes the problem computationally easier. We present a greedy algorithm to derive coverage-based prime implicant explanations when dealing with a tree-based model. We also present an empirical evaluation showing that this algorithm is efficient enough to be used in many practical cases.

We reformulate explanation quality assessment as a ranking problem rather than a generation problem. Instead of optimizing models to produce a single "best" explanation token-by-token, we train reward models to discriminate among multiple candidate explanations and learn their relative quality. Concretely, we construct per-instance candidate sets with graded quality levels and train listwise and pairwise ranking models (ListNet, LambdaRank, RankNet) to preserve ordinal structure and avoid score compression typical of pointwise regression or binary preference objectives. We observe three findings: First, ranking losses consistently outperform regression on score separation across all domains tested. Second, the optimal ranking loss depends on data characteristics: listwise objectives excel with well-separated quality tiers, while pairwise methods are more robust to noisy natural annotations. Third, when trained on carefully curated and well-structured data, small encoder models can match models that are orders of magnitude larger, suggesting that data quality matters more than model scale. Finally, when used as rewards in policy optimization, ranking-based scores enable stable convergence in settings where regression-based rewards fail entirely. Code and data are available at: https://github.com/Tankiit/PPO_Learning_to_rank

Concept Bottleneck Models (CBMs) predict through human-interpretable concepts, but they typically output point concept probabilities that conflate epistemic uncertainty (reducible model underspecification) with aleatoric uncertainty (irreducible input ambiguity). This makes concept-level uncertainty hard to interpret and, more importantly, hard to act upon. We introduce (Credal Ensemble Concept Estimation), a CBM framework that decomposes concept uncertainty by construction. represents each concept as a credal prediction (a probability interval), derives epistemic uncertainty from disagreement across diverse concept heads, and estimates aleatoric uncertainty via a dedicated ambiguity output trained to match annotator disagreement when available. The resulting signals support prescriptive decisions: automate low-uncertainty cases, prioritize data collection for high-epistemic cases, route high-aleatoric cases to human review, and abstain when both are high. Across several tasks, we show that epistemic uncertainty is positively associated with prediction errors, whereas aleatoric uncertainty closely tracks annotator disagreement, providing guidance beyond error correlation. Our implementation is available at the following link: https://github.com/Tankiit/Credal_Sets/tree/ensemble-credal-cbm

We present a new approach for distilling boosted trees into decision trees, in the objective of generating an ML model offering an acceptable compromise in terms of predictive performance and interpretability. We explain how the correction approach called rectification can be used to implement such a distillation process. We show empirically that this approach provides interesting results, in comparison with an approach to distillation achieved by retraining the model.

Predictive uncertainty is commonly decomposed into epistemic and aleatoric components, but standard decompositions often produce strongly correlated estimates because both quantities are derived from the same predictive distribution. We study an alternative design principle, structural separation, which assigns epistemic and aleatoric uncertainty to disjoint parameter paths trained with distinct supervision targets: reducible prediction error for epistemic uncertainty and persistent label ambiguity for aleatoric uncertainty. We instantiate this principle in two supervised latent-variable models, a Credal Concept Bottleneck Model and a credal Self-Explaining Neural Network, and prove a gradient-isolation result showing that the two uncertainty heads are not coupled through shared training gradients under the proposed parameterization. Across five ambiguity-aware benchmarks, structural separation substantially reduces epistemic-aleatoric correlation while preserving predictive performance. Further analyses show that aleatoric estimates track annotator- or corpus-derived ambiguity, while epistemic estimates are more sensitive to prediction error and data availability. These results suggest that supervised latent-variable architectures provide a practical route toward uncertainty estimates that are not merely decorrelated, but operationally distinguishable.

Due to increasing pressure on emergency departments, particularly driven by population aging, decision support tools are needed to improve patient prioritization. We propose an AI-based approach to identify patients at risk of critical deterioration at emergency admission using only routinely collected medical-administrative data. Our method relies exclusively on PMSI data, mainly diagnoses and procedures, which are widely available in hospital information systems. A boosted tree model is trained to predict deterioration while providing formal explanations. Despite the absence of physiological data, the model achieves strong predictive performance and offers interpretability. A subset of predictions reviewed by physicians did not reveal obvious errors.

We present Mode(Multi-Objective adaptive Data Efficiency), a framework that dynamically combines coreset selection strategies based on their evolving contribution to model performance. Unlike static methods, adapts selection criteria to training phases: emphasizing class balance early, diversity during representation learning, and uncertainty at convergence. We show that MODE achieves (1-1/e)-approximation with O(n logn) complexity and demonstrates competitive accuracy while providing interpretable insights into data utility evolution. Experiments show reduces memory requirements

In this work, we show how the class of improvement operators - a general class of iterated belief change operators - can be used to define a learning model. Focusing on binary classification, we present learning and inference algorithms suited to this learning model and we evaluate them empirically. Our findings highlight two key insights: first, that iterated belief change can be viewed as an effective form of online learning, and second, that the well-established axiomatic foundations of belief change operators offer a promising avenue for the axiomatic study of classification tasks.

We present a new approach to classification that combines data and knowledge. In this approach, data mining is used to derive association rules (possibly with negations) from data. Those rules are leveraged to increase the predictive performance of tree-based models (decision trees and random forests) used for a classification task. They are also used to improve the corresponding explanation task through the generation of abductive explanations that are more general than those derivable without taking such rules into account. Experiments show that for the two tree-based models under consideration, benefits can be offered by the approach in terms of predictive performance and in terms of explanation sizes.

In this article, we introduce a neuro-symbolic approach that combines a low-level perception task performed by a neural network with a high-level reasoning task performed by a possibilistic rule-based system. The goal is to be able to derive for each input instance the degree of possibility that it belongs to a target (meta-)concept. This (meta-)concept is connected to intermediate concepts by a possibilistic rule-based system. The probability of each intermediate concept for the input instance is inferred using a neural network. The connection between the low-level perception task and the high-level reasoning task lies in the transformation of neural network outputs modeled by probability distributions (through softmax activation) into possibility distributions. The use of intermediate concepts is valuable for the explanation purpose: using the rule-based system, the classification of an input instance as an element of the (meta-)concept can be justified by the fact that intermediate concepts have been recognized. From the technical side, our contribution consists of the design of efficient methods for defining the matrix relation and the equation system associated with a possibilistic rule-based system. The corresponding matrix and equation are key data structures used to perform inferences from a possibilistic rule-based system and to learn the values of the rule parameters in such a system according to a training data sample. Furthermore, leveraging recent results on the handling of inconsistent systems of fuzzy relational equations, an approach for learning rule parameters according to multiple training data samples is presented. Experiments carried out on the MNIST addition problems and the MNIST Sudoku puzzles problems highlight the effectiveness of our approach compared with state-of-the-art neuro-symbolic ones.

With the introduction of large language models, NLP has undergone a paradigm shift where these models now serve as the backbone of most developed systems. However, while highly effective, they remain opaque and difficult to interpret, which limits their adoption in critical applications that require transparency and trust. Two major approaches aim to address this: rationale extraction, which highlights input spans that justify predictions, and concept bottleneck models, which make decisions through human-interpretable concepts. Yet each has limitations-rationales lack semantic abstraction while concepts miss finegrained linguistic grounding. Crucially, current models lack a unified framework that connects where a model looks (rationales) with why it makes a decision (concepts). We introduce CLARITY, a model that first selects key input spans, maps them to interpretable concepts grounded in linguistic semantics, and then predicts using only those concepts. This design reveals how surface-level linguistic patterns map to abstract semantic representations, supporting faithful, multi-level explanations and allowing users to intervene at both the rationale and concept levels. CLARITY achieves competitive accuracy while offering improved transparency and semantic interpretability. The source code can be accessed at this link: CLARITY.

We present and evaluate empirically an XAI protocol for ruling interactions between a tree-based ML model (the AI system) and its user U, in the context of a prediction task. The pieces of knowledge held by U concerning the prediction task are supposed to be representable by a set of classification rules that is reliable and consistent, but (typically) incomplete. The proposed protocol aims to help U decide what to do with each prediction made by AI (accept it, reject it). It also aims to improve the quality of further predictions made by AI thanks to the expertise of U, and, reciprocally, to complete the pieces of knowledge held by U by leveraging the predictions made by AI. Experiments show that the approach can prove valuable in practice.

We show how to define and compute example-based abductive explanations. Such explanations are guaranteed to be 100% correct, fairly general, and persuasive enough since they cover sufficiently many reference instances furnished by the explainee. We prove that the latter coverage condition yields a complexity shift to the second level of the polynomial hierarchy. We present a CEGAR-based algorithm to derive such explanations, and show how to modify it to derive most anchored example-based abductive explanations, i.e., example-based abductive explanations that cover as many reference instances as possible. We also explain how to reduce example-based abductive explanations to get subset-minimal explanations. Experiments in the case of random forest classifiers show that our CEGAR-based algorithm is quite efficient in practice.

We are interested in identifying the complexity of computing local explanations of various types given a decision tree, when the Boolean conditions used in the tree are not independent. This is usually the case when decision trees are learned from instances described using numerical or categorical attributes. In such a case, considering the domain theory indicating how the Boolean conditions occurring in the tree are logically connected is paramount to derive provably correct explanations. However, the nature of the domain theory may have a strong impact on the complexity of generating explanations. In this paper, we identify the complexity of deriving local explanations (abductive or contrastive) given a decision tree in the general case, and under several natural restrictions about the domain theory.

The purpose of this book is to provide an overview of AI research, ranging from basic work to interfaces and applications, with as much emphasis on results as on current issues. It is aimed at an audience of master students and Ph.D. students, and can be of interest as well for researchers and engineers who want to know more about AI. The book is split into three volumes.

A contrastive explanation is a local explanation that is looked for when the prediction achieved by an ML model on an input instance x differs from what was foreseen. A contrastive explanation indicates how to change x to another instance x(c) from which a prediction that complies with the user's expectations can be obtained. In this paper, we present a constraint-based approach to the generation of contrastive explanations that are suited to regression functions represented by boosted trees. We show how to compute the smallest interval containing all the regression values that are attainable given a set of characteristics of x that are protected (i.e., not amenable to change). We also show how to generate minimal contrastive explanations for x given a target interval, i.e., instances with regression values within the specified interval and that are as close as possible to x. Closeness is captured using user-dependent mappings reflecting preferences about value change for the attributes (or combinations of attributes) considered in the representation of x.

In this paper, we explore the application of blocked clause elimination for projected model counting. This is the problem of determining the number of models ||\exists X.{\Sigma}|| of a propositional formula {\Sigma} after eliminating a given set X of variables existentially. Although blocked clause elimination is a well-known technique for SAT solving, its direct application to model counting is challenging as in general it changes the number of models. However, we demonstrate, by focusing on projected variables during the blocked clause search, that blocked clause elimination can be leveraged while preserving the correct model count. To take advantage of blocked clause elimination in an efficient way during model counting, a novel data structure and associated algorithms are introduced. Our proposed approach is implemented in the model counter d4. Our experiments demonstrate the computational benefits of our new method of blocked clause elimination for projected model counting.

In a recently published book, the French writer and comedian François Rollin has discussed various aspects of the notion of stupidity, including artificial stupidity, the stupid counterpart of artificial intelligence. His claim is that a system of artificial stupidity is a system that provides wrong answers to any task it should solve, leading to absurd solutions in most cases. We believe that this claim is (at least partially) false and that designing artificial stupidity is not as trivial as it seems. In this article, we discuss why and how one could design a system of artificial stupidity. We believe that such a reflection on (artificial) stupidity can bring about some interesting insights about (artificial) intelligence.