Committed research and published works.
CORTEX, RESTORE and TRACEFORM-AI start in 2026. Existing publications concern, among others, LAMMR, presented at HHAI 2026, and BRAIN BiasAI.
Integrity, innovation, impact, collaboration, agility.
Five values guide our research and collaborations: ethical, transparent and accountable AI; curiosity; results that are useful to clients and society; co-creation with industry and academia; the ability to learn fast.
A hybrid governance model.
An internal scientific board and an external advisory network of academic and industry experts guide the research. Aligned with DTSC’s mission, the laboratory retains its research autonomy for its academic collaborations and experiments. It takes part in joint PhDs, public-private partnerships and competitive research programmes, such as Win4Doc, Horizon Europe or the EDIHs.
Three complementary pillars.
- DTSC’s investment, which supports the laboratory’s direction, its early research, its infrastructure and its talent.
- Public funding, regional, national and European: Win4Doc and Win4Company in Wallonia, Innoviris in Brussels, Baekeland mandates in Flanders, and European frameworks.
- Collaborations with companies : industrial partnerships, contract research and applied AI projects.
Work accessible to as many people as possible.
Depending on the project, we share our work through open-access publications, reproducible code, public datasets and knowledge-sharing events.
Scientific publications
- Learning and Reasoning for Legal AI via Intelligent Adaptive Multi-Model Routing with a Smart Evaluation LayerAditya Mukhopadhyay, Sara Ghane, Marco Di GennaroHHAI 2026: Proceedings of the 5th International Conference on Hybrid Human-Artificial Intelligence. SAGE Publications, 2026
- Benchmarking Physics-Injection Strategies for Transformer Hot-Spot Prediction: Residual, Physics-Informed, and Dynamic Gray-Box ModelsMario Real Enrique, Enrique Hortal, Sara Ghane, Marco Di Gennaro61st International Universities Power Engineering Conference, 2026
- Assessing railway network robustness under attacks: Evidence from Belgium and the NetherlandsPraneesh Sharma, Renzo Massobrio, Marco Di GennaroPhysica A: Statistical Mechanics and its Applications, 2026
- When Corporate Chatbots Show Bias: A Multi-Dimensional Analysis of LLMs in Enterprise SettingsShreya Bhattacharya, Vincent Hagenow, Marco Di GennaroECAI, 2025
Master theses
- Physics-Informed Machine Learning for Transformer Thermal State Estimation: A Comparative StudyMario Real EnriqueMaastricht University, Netherlands
- Residential Energy Advising with Small Language Models: A Tool-Calling Architecture and Benchmark EvaluationM. Turan TahsinKU Leuven, Belgium
- Measuring Grammatical CompetenceAditya MukhopadhyayManipal Academy of Higher Education, India
- Toward Transparent Fault Diagnosis in Grid-Interactive Building Energy Control: Using SHAP and LLMs to Explain Black-Box ControllersEdouard SteylemanKU Leuven, Belgium
- Digital Twin for railway network simulationEthan BebelambouUniversité libre de Bruxelles (ULB), Belgium
Source: the laboratory’s “BRAIN projects” document, version of 13 November 2026. References are given in their original language.
Move the ecosystem forward.
- Scientific and academic cooperations
- Research on collective issues
- Transfer and sharing of results according to the project