Marie-Jeanne Lesot

Milan Bhan, Jean-Noel Vittaut, Nicolas Chesneau et al.

Counterfactual examples explain a prediction by highlighting changes of instance that flip the outcome of a classifier. This paper proposes TIGTEC, an efficient and modular method for generating sparse, plausible and diverse counterfactual explanations for textual data. TIGTEC is a text editing heuristic that targets and modifies words with high contribution using local feature importance. A new attention-based local feature importance is proposed. Counterfactual candidates are generated and assessed with a cost function integrating semantic distance, while the solution space is efficiently explored in a beam search fashion. The conducted experiments show the relevance of TIGTEC in terms of success rate, sparsity, diversity and plausibility. This method can be used in both model-specific or model-agnostic way, which makes it very convenient for generating counterfactual explanations.

Adulam Jeyasothy, Thibault Laugel, Marie-Jeanne Lesot et al.

In the field of eXplainable Artificial Intelligence (XAI), post-hoc interpretability methods aim at explaining to a user the predictions of a trained decision model. Integrating prior knowledge into such interpretability methods aims at improving the explanation understandability and allowing for personalised explanations adapted to each user. In this paper, we propose to define a cost function that explicitly integrates prior knowledge into the interpretability objectives: we present a general framework for the optimization problem of post-hoc interpretability methods, and show that user knowledge can thus be integrated to any method by adding a compatibility term in the cost function. We instantiate the proposed formalization in the case of counterfactual explanations and propose a new interpretability method called Knowledge Integration in Counterfactual Explanation (KICE) to optimize it. The paper performs an experimental study on several benchmark data sets to characterize the counterfactual instances generated by KICE, as compared to reference methods.

Nour Bouchouchi, Thiabult Laugel, Xavier Renard et al.

During training, Large Language Models (LLMs) learn social regularities that can lead to gender bias in downstream applications. Most mitigation efforts focus on reducing bias in generated outputs, typically evaluated on structured benchmarks, which raises two concerns: output-level evaluation does not reveal whether alignment modifies the model's underlying representations, and structured benchmarks may not reflect realistic usage scenarios. We propose a unified framework to jointly analyze intrinsic and extrinsic gender bias in LLMs using identical neutral prompts, enabling direct comparison between gender-related information encoded in internal representations and bias expressed in generated outputs. Contrary to prior work reporting weak or inconsistent correlations, we find a consistent association between latent gender information and expressed bias when measured under the unified protocol. We further examine the effect of alignment through supervised fine-tuning aimed at reducing gender bias. Our results suggest that while the latter indeed reduces expressed bias, measurable gender-related associations are still present in internal representations, and can be reactivated under adversarial prompting. Finally, we consider two realistic settings and show that debiasing effects observed on structured benchmarks do not necessarily generalize, e.g., to the case of story generation.

Adam Rida, Marie-Jeanne Lesot, Xavier Renard et al.

Explainable AI (XAI) methods have mostly been built to investigate and shed light on single machine learning models and are not designed to capture and explain differences between multiple models effectively. This paper addresses the challenge of understanding and explaining differences between machine learning models, which is crucial for model selection, monitoring and lifecycle management in real-world applications. We propose DeltaXplainer, a model-agnostic method for generating rule-based explanations describing the differences between two binary classifiers. To assess the effectiveness of DeltaXplainer, we conduct experiments on synthetic and real-world datasets, covering various model comparison scenarios involving different types of concept drift.

Yann Munro, Camilo Sarmiento, Isabelle Bloch et al.

An abstract argumentation framework is a commonly used formalism to provide a static representation of a dialogue. However, the order of enunciation of the arguments in an argumentative dialogue is very important and can affect the outcome of this dialogue. In this paper, we propose a new framework for modelling abstract argumentation graphs, a model that incorporates the order of enunciation of arguments. By taking this order into account, we have the means to deduce a unique outcome for each dialogue, called an extension. We also establish several properties, such as termination and correctness, and discuss two notions of completeness. In particular, we propose a modification of the previous transformation based on a "last enunciated last updated" strategy, which verifies the second form of completeness.

Yann Munro, Isabelle Bloch, Marie-Jeanne Lesot

Formal argumentation is being used increasingly in artificial intelligence as an effective and understandable way to model potentially conflicting pieces of information, called arguments, and identify so-called acceptable arguments depending on a chosen semantics. This paper deals with the specific context of Quantitative Bipolar Argumentation Frameworks (QBAF), where arguments have intrinsic weights and can attack or support each other. In this context, we introduce a novel family of gradual semantics, called aggregative semantics. In order to deal with situations in which attackers and supporters do not play a symmetric role, and in contrast to modular semantics, we propose to aggregate attackers and supporters separately. This leads to a three-stage computation, which consists in computing a global weight for attackers and another for supporters, before aggregating these two values with the intrinsic weight of the argument. We discuss the properties that the three aggregation functions should satisfy depending on the context, as well as their relationships with the classical principles for gradual semantics. This discussion is supported by various simple examples, as well as a final example on which five hundred aggregative semantics are tested and compared, illustrating the range of possible behaviours with aggregative semantics. Decomposing the computation into three distinct and interpretable steps leads to a more parametrisable computation: it keeps the bipolarity one step further than what is done in the literature, and it leads to more understandable gradual semantics.

Milan Bhan, Jean-Noel Vittaut, Nicolas Chesneau et al.

Incorporating natural language rationales in the prompt and In-Context Learning (ICL) have led to a significant improvement of Large Language Models (LLMs) performance. However, generating high-quality rationales require human-annotation or the use of auxiliary proxy models. In this work, we propose Self-AMPLIFY to automatically generate rationales from post hoc explanation methods applied to Small Language Models (SLMs) to improve their own performance. Self-AMPLIFY is a 3-step method that targets samples, generates rationales and builds a final prompt to leverage ICL. Self-AMPLIFY performance is evaluated on four SLMs and five datasets requiring strong reasoning abilities. Self-AMPLIFY achieves good results against competitors, leading to strong accuracy improvement. Self-AMPLIFY is the first method to apply post hoc explanation methods to autoregressive language models to generate rationales to improve their own performance in a fully automated manner.

Clara Bove, Thibault Laugel, Marie-Jeanne Lesot et al.

As Machine Learning models achieve unprecedented levels of performance, the XAI domain aims at making these models understandable by presenting end-users with intelligible explanations. Yet, some existing XAI approaches fail to meet expectations: several issues have been reported in the literature, generally pointing out either technical limitations or misinterpretations by users. In this paper, we argue that the resulting harms arise from a complex overlap of multiple failures in XAI, which existing ad-hoc studies fail to capture. This work therefore advocates for a holistic perspective, presenting a systematic investigation of limitations of current XAI methods and their impact on the interpretation of explanations. % By distinguishing between system-specific and user-specific failures, we propose a typological framework that helps revealing the nuanced complexities of explanation failures. Leveraging this typology, we discuss some research directions to help practitioners better understand the limitations of XAI systems and enhance the quality of ML explanations.

Milan Bhan, Jean-Noel Vittaut, Nina Achache et al.

Toxicity mitigation consists in rephrasing text in order to remove offensive or harmful meaning. Neural natural language processing (NLP) models have been widely used to target and mitigate textual toxicity. However, existing methods fail to detoxify text while preserving the initial non-toxic meaning at the same time. In this work, we propose to apply counterfactual generation methods from the eXplainable AI (XAI) field to target and mitigate textual toxicity. In particular, we perform text detoxification by applying local feature importance and counterfactual generation methods to a toxicity classifier distinguishing between toxic and non-toxic texts. We carry out text detoxification through counterfactual generation on three datasets and compare our approach to three competitors. Automatic and human evaluations show that recently developed NLP counterfactual generators can mitigate toxicity accurately while better preserving the meaning of the initial text as compared to classical detoxification methods. Finally, we take a step back from using automated detoxification tools, and discuss how to manage the polysemous nature of toxicity and the risk of malicious use of detoxification tools. This work is the first to bridge the gap between counterfactual generation and text detoxification and paves the way towards more practical application of XAI methods.

Milan Bhan, Yann Choho, Pierre Moreau et al.

Textual Concept Bottleneck Models (TCBMs) are interpretable-by-design models for text classification that predict a set of salient concepts before making the final prediction. This paper proposes Complete Textual Concept Bottleneck Model (CT-CBM), a novel TCBM generator building concept labels in a fully unsupervised manner using a small language model, eliminating both the need for predefined human labeled concepts and LLM annotations. CT-CBM iteratively targets and adds important and identifiable concepts in the bottleneck layer to create a complete concept basis. CT-CBM achieves striking results against competitors in terms of concept basis completeness and concept detection accuracy, offering a promising solution to reliably enhance interpretability of NLP classifiers.

Ekaterina Goliakova, Xavier Renard, Marie-Jeanne Lesot et al.

Using multiple-choice questions (MCQs) has become a standard for assessing LLM capabilities efficiently. A variety of metrics can be employed for this task. However, previous research has not conducted a thorough assessment of them. At the same time, MCQ evaluation suffers from answer fluctuation: models produce different results given slight changes in prompts. We suggest a metric assessment protocol in which evaluation methodologies are analyzed through their connection with fluctuation rates, as well as original performance. Our results show that there is a strong link between existing metrics and the answer changing, even when computed without any additional prompt variants. A novel metric, worst accuracy, demonstrates the highest association on the protocol.

Milan Bhan, Jean-Noel Vittaut, Nicolas Chesneau et al.

Large Language Models (LLMs) can generate plausible free text self-explanations to justify their answers. However, these natural language explanations may not accurately reflect the model's actual reasoning process, indicating a lack of faithfulness. Existing faithfulness evaluation methods rely primarily on behavioral tests or computational block analysis without examining the semantic content of internal neural representations. This paper proposes NeuroFaith, a flexible framework that measures the faithfulness of LLM free text self-explanation by identifying key concepts within explanations and mechanistically testing whether these concepts actually influence the model's predictions. We show the versatility of NeuroFaith across 2-hop reasoning and classification tasks. Additionally, a linear faithfulness probe based on NeuroFaith is developed to detect unfaithful self-explanations from representation space and improve faithfulness through steering. NeuroFaith provides a principled approach to evaluating and enhancing the faithfulness of LLM free text self-explanations, addressing critical needs for trustworthy AI systems.

Thibault Laugel, Adulam Jeyasothy, Marie-Jeanne Lesot et al.

In the field of Explainable Artificial Intelligence (XAI), counterfactual examples explain to a user the predictions of a trained decision model by indicating the modifications to be made to the instance so as to change its associated prediction. These counterfactual examples are generally defined as solutions to an optimization problem whose cost function combines several criteria that quantify desiderata for a good explanation meeting user needs. A large variety of such appropriate properties can be considered, as the user needs are generally unknown and differ from one user to another; their selection and formalization is difficult. To circumvent this issue, several approaches propose to generate, rather than a single one, a set of diverse counterfactual examples to explain a prediction. This paper proposes a review of the numerous, sometimes conflicting, definitions that have been proposed for this notion of diversity. It discusses their underlying principles as well as the hypotheses on the user needs they rely on and proposes to categorize them along several dimensions (explicit vs implicit, universe in which they are defined, level at which they apply), leading to the identification of further research challenges on this topic.

Adam Faci, Marie-Jeanne Lesot, Claire Laudy

Conceptual Graphs (CG) are a graph-based knowledge representation and reasoning formalism; fuzzy Conceptual Graphs (fCG) constitute an extension that enriches their expressiveness, exploiting the fuzzy set theory so as to relax their constraints at various levels. This paper proposes a comparative study of existing approaches over their respective advantages and possible limitations. The discussion revolves around three axes: (a) Critical view of each approach and comparison with previous propositions from the state of the art; (b) Presentation of the many possible interpretations of each definition to illustrate its potential and its limits; (c) Clarification of the part of CG impacted by the definition as well as the relaxed constraint.

Adam Faci, Marie-Jeanne Lesot, Claire Laudy

Conceptual Graphs (CGs) are a graph-based knowledge representation formalism. In this paper we propose cgSpan a CG frequent pattern mining algorithm. It extends the DMGM-GSM algorithm that takes taxonomy-based labeled graphs as input; it includes three more kinds of knowledge of the CG formalism: (a) the fixed arity of relation nodes, handling graphs of neighborhoods centered on relations rather than graphs of nodes, (b) the signatures, avoiding patterns with concept types more general than the maximal types specified in signatures and (c) the inference rules, applying them during the pattern mining process. The experimental study highlights that cgSpan is a functional CG Frequent Pattern Mining algorithm and that including CGs specificities results in a faster algorithm with more expressive results and less redundancy with vocabulary.

Thibault Laugel, Marie-Jeanne Lesot, Christophe Marsala et al.

Post-hoc interpretability approaches have been proven to be powerful tools to generate explanations for the predictions made by a trained black-box model. However, they create the risk of having explanations that are a result of some artifacts learned by the model instead of actual knowledge from the data. This paper focuses on the case of counterfactual explanations and asks whether the generated instances can be justified, i.e. continuously connected to some ground-truth data. We evaluate the risk of generating unjustified counterfactual examples by investigating the local neighborhoods of instances whose predictions are to be explained and show that this risk is quite high for several datasets. Furthermore, we show that most state of the art approaches do not differentiate justified from unjustified counterfactual examples, leading to less useful explanations.

Thibault Laugel, Marie-Jeanne Lesot, Christophe Marsala et al.

Counterfactual post-hoc interpretability approaches have been proven to be useful tools to generate explanations for the predictions of a trained blackbox classifier. However, the assumptions they make about the data and the classifier make them unreliable in many contexts. In this paper, we discuss three desirable properties and approaches to quantify them: proximity, connectedness and stability. In addition, we illustrate that there is a risk for post-hoc counterfactual approaches to not satisfy these properties.

Xavier Renard, Thibault Laugel, Marie-Jeanne Lesot et al.

Machine learning models are increasingly used in the industry to make decisions such as credit insurance approval. Some people may be tempted to manipulate specific variables, such as the age or the salary, in order to get better chances of approval. In this ongoing work, we propose to discuss, with a first proposition, the issue of detecting a potential local adversarial example on classical tabular data by providing to a human expert the locally critical features for the classifier's decision, in order to control the provided information and avoid a fraud.

Thibault Laugel, Xavier Renard, Marie-Jeanne Lesot et al.

Local surrogate models, to approximate the local decision boundary of a black-box classifier, constitute one approach to generate explanations for the rationale behind an individual prediction made by the back-box. This paper highlights the importance of defining the right locality, the neighborhood on which a local surrogate is trained, in order to approximate accurately the local black-box decision boundary. Unfortunately, as shown in this paper, this issue is not only a parameter or sampling distribution challenge and has a major impact on the relevance and quality of the approximation of the local black-box decision boundary and thus on the meaning and accuracy of the generated explanation. To overcome the identified problems, quantified with an adapted measure and procedure, we propose to generate surrogate-based explanations for individual predictions based on a sampling centered on particular place of the decision boundary, relevant for the prediction to be explained, rather than on the prediction itself as it is classically done. We evaluate the novel approach compared to state-of-the-art methods and a straightforward improvement thereof on four UCI datasets.

Thibault Laugel, Marie-Jeanne Lesot, Christophe Marsala et al.

In the context of post-hoc interpretability, this paper addresses the task of explaining the prediction of a classifier, considering the case where no information is available, neither on the classifier itself, nor on the processed data (neither the training nor the test data). It proposes an instance-based approach whose principle consists in determining the minimal changes needed to alter a prediction: given a data point whose classification must be explained, the proposed method consists in identifying a close neighbour classified differently, where the closeness definition integrates a sparsity constraint. This principle is implemented using observation generation in the Growing Spheres algorithm. Experimental results on two datasets illustrate the relevance of the proposed approach that can be used to gain knowledge about the classifier.