Search results

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, Japan

Can we still speak of democracy when algorithms influence our electoral choices or participate in the drafting of laws? This topic is explored by Aline Nardi, researcher at the Faculty of Law and member of the Namur Digital Institute (NADI).

The Namur Digital Institute (NaDI) is launching a series of original events: "Les Séances du Numérique". Films followed by debates with experts to understand digital challenges and stimulate collective thinking. A project spearheaded by Anthony Simonofski, Anne-Sophie Collard, Benoît Vanderose and Fanny Barnabé.

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, Japan

Welcome to the 1ʳᵉ MG-ERC conference This conference, linked to the research themes of the Chemistry Department, aims to bring together around 100 researchers working in the fields of heteroatom chemistry, coordination chemistry, catalysis, and inorganic chemistry. It represents a real novelty in Belgium in terms of the areas covered, and will enable participants to discover new concepts, ideas and trends in these recent areas of research in chemistry. Here is the list of speakers, who are world experts in their fieldsDr. Daniël Broere (Utrecht University, Netherlands)Prof. Agnieszka Nowak-Król (Universität Würzburg, Germany)Dr. Antoine Simonneau (Université Paul-Sabatier, Toulouse, France)Prof. Dr. Sebastian Riedel (Freie Universität, Berlin, Germany)Dr. Arnaud Voituriez (Université Paris-Saclay, France)Prof. Dr. Alessandro Bismuto (Universität Bonn, Germany)Dr. Christian Hering-Junghans (Leibniz-Institut für Katalyse, Germany)Prof. Connie Lu (Universität Bonn, Germany)Prof. Simon Aldridge (University of Oxford, UK)Dr. Ghenwa Bouhadir (Université Paul-Sabatier, Toulouse, France)Prof. Dr. Viktoria Däschlein (Universität Bonn, Germany)Prof. Viktoria Däschlein-Gessner (Ruhr-University of Bochum, Germany)Dr. Jennifer A. Garden (University of Edinburgh, UK)Prof. Muriel Hissler (Université de Rennes, France)Prof. Jean-François Paquin (Université de Laval, Canada) More information and registration

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, Japan

This is a great recognition of research at UNamur: three Marie Skłodowska-Curie Doctoral Networks (DN) projects have just been awarded, with a key contribution from researchers in Namur! The first, in chemistry, involves Professor Stéphane Vincent; the second, focused on ecosystem resilience, involves Professor Frédérik de Laender; and the third, in the field of photonics, benefits from the expertise of FNRS-qualified researcher Michaël Lobet.

Flamure Ibrahimi is a Ph.D. student in service and marketing management at the NaDI-CeRCLe Research Center at the University of Namur (Belgium) within the EMCP Faculty, under the supervision of Prof. Dr. Wafa Hammedi (University of Namur) and Prof. Dr. Linda Alkire (Texas State University). She has just been awarded the prestigious ServCollab Scholarship 2026, an international distinction that recognizes and supports doctoral students whose work falls within the field of Transformative Service Research (TSR)—doctoral projects with high impact on society and humanity.