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

Something for everyone 09:30 | Welcome ceremony for new students11:00 | Back-to-school celebration at Saint-Aubain Cathedral (Place Saint-Aubain - 5000 Namur), followed by student welcome by the Cercles. Read more

BNAIC/BeNeLearn 2025 will be held at the University of Namur under the auspices of the Belgian-Dutch Association for Artificial Intelligence (BNVKI) and the Dutch Research School for Information and Knowledge Systems (SIKS). The conference aims at presenting an overview of state-of-the-art research in artificial intelligence and machine learning in Belgium, The Netherlands, and Luxembourg. 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

A collaboration between UNamur - Hénallux - IFEC Image Image Image Since September 2022, Hénallux and UNamur have been collaborating with IFEC (Institut de Formation de l'Enseignement Catholique) with the aim of raising awareness and training those involved in secondary education in educational differentiation, advocated in the Pacte pour un enseignement d'excellence. To achieve this, various modalities have been devised and implemented: in-school pedagogical days, follow-ups in schools that request them, courses for differentiation referent teachers, courses for CSAs (support and accompaniment advisors).The convention is coordinated by Sandrine Biémar (UNamur) and Alain Bultot (Hénallux), and the team of researchers includes Anne Libert (UNamur), Virginie Meyer (UNamur) and Sylvie Van der Linden (Hénallux). Background This collaborative research is part of a wider project devoted to differentiation, which brings together several courses with different aims and audiences. These include in-school pedagogical training, the training of teacher-referents whose vocation is to be facilitators of the development of differentiation practices in their schools, training for all IFEC CSAs, as well as regular meetings with IFEC in-house trainers interested in this theme. Collaborative research is part of this continuity, with each path nourishing the others. It aims to anchor reflection on differentiation in classroom practices in a mutual enrichment of theory and practice. In practice School courses : pedagogical training and follow-up courses lasting from 1 to several days. CSA course : 4-day training for all CSAs and IFEC. Referent courses : Training of pairs of school referents for 4 days + 1 day. Trainers' courses : community of practice with IFEC internal differentiation trainers Collaborative research path : accompanying teachers over a school year in implementing differentiation practices. The objectives Fostering teachers' power to act by mobilizing and interpreting objective data gathered in the field.Collaborative research is built on negotiation between participants and researchers. Each stage of the research is therefore constructed together, in order to respond as closely as possible to the concerns and issues related to differentiation that are encountered in the specific contexts of each participant.For example, the research question and sub-questions are co-formulated to respond as closely as possible to the concerns and issues in the field regarding the impact of a differentiated teaching posture on student motivation and autonomy.. Methodology This research is inspired by the protocol developed by Schildkamp (2018, 2019) within "Data TEAMS". It aims to develop and foster teachers' power to act through decision-making based on data collection and evaluation of school practices.The data resulting from this research pathway will feed the referent and trainer pathways. In addition, data and productions resulting from this collaborative research will also be destined for colleagues and organizations from the various stakeholders (schools, CSA). Want to get involved? We are looking for pairs of teachers from the same school, which will facilitate the implementation of the process within each school concerned. The themeDevelop and regulate differentiation practices in light of classroom data.The objectivesUnderstand and actUnderstand, through analysis of available classroom data and exchanges of practice between professionals, the effects of a differentiated teacher posture on student motivation and autonomy.Objectivate one's intuition to act effectively.Terms and conditions7 meetings over the course of the year at the Salle des Pros (Rue Godefroid, 7 - in the center of Namur)Why participate?To enrich research by drawing on practices in the field.To enrich your practices thanks to the support of the researcher-trainers.To meet other teachers who share your concerns. Contact For further information, please contact the research teamAnne Libert : anne.libert@unamur.beVirginie Meyer : virginie.meyer@unamur.beSylvie Van der Linden : sylvie.vanderlinden@henallux.be Project coordinator Sandrine Biémar: sandrine.biemar@unamur.be

This is great recognition for the excellence of the research carried out at the University of Namur: the Faculty of Informatics has been asked to join the prestigious Informatics Europe network, which brings together the most dynamic departments and faculties of Informatics across Europe.

Critical thinking, the art of productive doubt, can be learned and cultivated. Faced with information overload and the spread of artificial intelligence, it is more important than ever for students to develop this skill throughout their studies. At UNamur, this educational necessity takes many forms. 

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

Japan is almost 10,000 kilometers from Belgium, a country that fascinates, not least for its rich culture full of contrasts. Researchers at UNamur maintain close ties with several Japanese institutions, particularly in the fields of computer science, mathematics and video games. Let's take a look at some of these collaborations..

FaSEF provides theoretical and practical training in preparation for the CAPAES (Certificate of Aptitude for Teaching in Higher Education) for teachers employed under contract in a university or in higher education for social advancement outside the teaching categories.

The Master's degree specializing in coaching professionals in education, management, health, and social work (MAPEMASS) offers unique training and aims to provide trainers, team leaders, advisors, HR managers, and others with training in coaching teams and the professionals who comprise them.