Résultats de recherche

Abstract Detecting semantic code clones in logic programs is a longstanding challenge, due to the lack of a unified definition of semantic similarity and the diversity of syntactic expressions that can represent similar behaviours. This thesis introduces a formal and flexible framework for semantic clone detection based on Constrained Horn Clauses (CHC). The approach considers two predicates as semantic clones if they can be independently transformed, via semantics-preserving program transformations, into a common third predicate. At the core of the method lies anti-unification, a process that computes the most specific generalisation of two predicates by identifying their shared structural patterns. The framework is parametric in regard with the allowed program transformations, the notion of generality, and the so-called quality estimators that steer the anti-unification process. Jury Prof. Wim Vanhoof - University of Namur, BelgiumProf. Katrien Beuls - University of Namur, BelgiumProf. Jean-Marie Jacquet - University of Namur, BelgiumProf. Temur Kutsia - Johannes Kepler University, AustriaProf. Frédéric Mesnard - University of the Reunion, Reunion IslandProf. Paul Van Eecke - Free University of Brussels, Belgium

Abstract In recent decades, the volume of data generated worldwide has grown exponentially, significantly accelerating advancements in machine learning. This explosion of data has led to an increased need for effective data exploration techniques, giving rise to a specialized field known as dimensionality reduction. Dimensionality reduction methods are used to transform high-dimensional data into a low-dimensional space (typically 2D or 3D), so that it can be easily visualized and understood by humans. Algorithms such as Principal Component Analysis (PCA), Multidimensional Scaling (MDS), and t-distributed Stochastic Neighbor Embedding (t-SNE) have become essential tools for visualizing complex datasets. These techniques play a critical role in exploratory data analysis and in interpreting complex models like Convolutional Neural Networks (CNNs). Despite their widespread adoption, dimensionality reduction techniques, particularly non-linear ones, often lack interpretability. This opacity makes it difficult for users to understand the meaning of the visualizations or the rationale behind specific low-dimensional representations. In contrast, the field of supervised machine learning has seen significant progress in explainable AI (XAI), which aims to clarify model decisions, especially in high-stakes scenarios. While many post-hoc explanation tools have been developed to interpret the outputs of supervised models, there is still a notable gap in methods for explaining the results of dimensionality reduction techniques. This research investigates how post-hoc explanation techniques can be integrated into dimensionality reduction algorithms to improve user understanding of the resulting visualizations. Specifically, it explores how interpretability methods originally developed for supervised learning can be adapted to explain the behavior of non-linear dimensionality reduction algorithms. Additionally, this work examines whether the integration of post-hoc explanations can enhance the overall effectiveness of data exploration. As these tools are intended for end-users, we also design and evaluate an interactive system that incorporates explanatory mechanisms. We argue that combining interpretability with interactivity significantly improves users' understanding of embeddings produced by non-linear dimensionality reduction techniques. In this research, we propose enhancements to an existing post-hoc explanation method that adapts LIME for t-SNE. We introduce a globally-local framework for fast and scalable explanations of t-SNE embeddings. Furthermore, we present a completely new approach that adapts saliency map-based explanations to locally interpret non-linear dimensionality reduction results. Lastly, we introduce our interactive tool, Insight-SNE, which integrates our gradient-based explanation method and enables users to explore low-dimensional embeddings through direct interaction with the explanations. Jury Prof. Wim Vanhoof - University of Namur, BelgiumProf. Benoit Frénay - University of Namur, BelgiumProf. Bruno Dumas - University of Namur, BelgiumProf. John Lee - University of Louvain, BelgiumProf. Luis Galarraga - University of Rennes, France

Près de 10 000 kilomètres séparent la Belgique du Japon, un pays qui fascine, notamment pour sa culture riche et pleine de contrastes. Les chercheurs de l’UNamur entretiennent des liens étroits avec plusieurs institutions nipponnes, notamment dans les domaines de l’informatique, des mathématiques ou encore du jeu vidéo. Coup de projecteur sur quelques-unes de ces collaborations.

Abstract Deep learning has become an extremely important technology in numerous domains such as computer vision, natural language processing, and autonomous systems. As neural networks grow in size and complexity to meet the demands of these applications, the cost of designing and training efficient models continues to rise in computation and energy consumption. Neural Architecture Search (NAS) has emerged as a promising solution to automate the design of performant neural networks. However, conventional NAS methods often require evaluating thousands of architectures, making them extremely resource-intensive and environmentally costly.This thesis introduces a novel, energy-aware NAS pipeline that operates at the intersection of Software Engineering and Machine Learning. We present CNNGen, a domain-specific generator for convolutional architectures, combined with performance and energy predictors to drastically reduce the number of architectures that need full training. These predictors are integrated into a multi-objective genetic algorithm (NSGA-II), enabling an efficient search for architectures that balance accuracy and energy consumption.Our approach explores a variety of prediction strategies, including sequence-based models, image-based representations, and deep metric learning, to estimate model quality from partial or symbolic representations. We validate our framework across three benchmark datasets, CIFAR-10, CIFAR-100, and Fashion-MNIST, demonstrating that it can produce results comparable to state-of-the-art architectures with significantly lower computational cost. By reducing the environmental footprint of NAS while maintaining high performance, this work contributes to the growing field of Green AI and highlights the value of predictive modelling in scalable and sustainable deep learning workflows. Jury Prof. Wim Vanhoof - University of Namur, BelgiumProf. Gilles Perrouin - University of Namur, BelgiumProf. Benoit Frénay - University of Namur, BelgiumProf. Pierre-Yves Schobbens - University of Namur, BelgiumProf. Clément Quinton - University of Lille, FranceProf. Paul Temple- University of Rennes, FranceProf. Schin’ichi Satoh - National Institute of Informatics, Japon

Abstract Plastic pollution has emerged as a pervasive environmental threat, with micro- and nanoplastics (MPs and NPs) accumulating across ecosystems and organisms, including humans. Their ability to adsorb and transport contaminants raises critical concerns for both environmental and public health.This thesis investigates the developmental neurotoxicity of MPs and NPs in zebrafish (Danio rerio), emphasizing the influence of particle size and mixture toxicity. NPs were shown to cross the embryonic chorion, disrupt physiological functions, and induce anxiety-like behaviour, whereas MPs mainly altered gene expression related to neurodevelopment. When co-exposed with methylmercury (MeHg), NPs enhanced MeHg accumulation in the brain and sensory organs, exacerbating its neurotoxic effects. Notably, the mixture induced severe hypoactivity, impaired lipid metabolism and neurotransmission, and increased mortality.These findings highlight the critical need to assess plastic particle toxicity not only in isolation but also in environmentally relevant mixtures. NPs, due to their small size and high reactivity, may act as vectors for toxicants like MeHg, amplifying their effects during sensitive developmental stages. Jury Prof. Frédéric SILVESTRE (UNamur), PrésidentProf. Patrick KESTEMONT (UNamur), SecrétaireDr Valérie CORNET (UNamur)Prof. Eli THORÉ (UNamur)Prof. Jérôme CACHOT (Université de Bordeaux)Dr Krishna DAS (Université de Liège)

The workshop will feature:A keynote presentation on public value and AI implementation at VRT.Sessions on discoverability, user agency, and explainability.Discussions on regulation, including perspectives on the AI Act and transparency in media.An interactive session showcasing AI-driven prototypes.The event will also highlight our project’s latest findings. Join us for a day of thought-provoking discussions, knowledge exchange, and networking opportunities!Would you like to attend? Places are limited and will be allocated on a first-come, first-served basis, so register as soon as possible. Registration will close on April 11, 2025. More information here

Un ordinateur peut-il apprendre une langue comme le fait un enfant ? Une étude récente publiée dans la revue de référence Computational Linguistics par les professeurs Katrien Beuls (Université de Namur) et Paul Van Eecke (AI-lab, Vrije Universiteit Brussel) apporte un nouvel éclairage sur cette question. Les chercheurs plaident pour une révision fondamentale de la manière dont l'intelligence artificielle acquiert et traite le langage. 

Terminator pour parler d’IA ? Wall-E pour parler de la dépendance technologique ? The Truman Show pour évoquer les réseaux sociaux ? Dans un nouvel ouvrage, deux professeurs de l’UNamur, Anthony Simonofski (transformation numérique- Faculté EMCP – Institut NaDI) et Benoît Vanderose (Génie logiciel – Faculté d’informatique – Institut NaDI), proposent un voyage à la croisée du numérique et de l’imaginaire cinématographique.