Thomas Demeester

Thomas Demeester is an Associate Professor at the Internet Technology and Data Science Lab (IDLab), Ghent University - imec, Belgium. Appointed as Assistant Professor in 2019, he leads an AI research group focused on health applications and drug design, co-directing the Text-to-Knowledge research cluster with Prof. Chris Develder.

His educational background includes:

• M.Sc. in Electrical Engineering from Ghent University (2005), completed with thesis work at ETH Zurich
• Ph.D. in Computational Electromagnetics from Ghent University (2009), funded by Research Foundation - Flanders (FWO)

Demeester's research spans artificial intelligence with emphasis on deep learning and neuro-symbolic methods. Current tracks include energy-based models (Hopfield Networks, Deep Equilibrium Models), diffusion models for drug design, and clinical reasoning systems. His work bridges NLP, healthcare informatics, and generative AI with strong industry partnerships.

Recent publications (2023-2025) reveal strategic expansion from NLP into health-centric AI: BioLORD biomedical encoders (2023), synthetic medical data frameworks (UAI/NeurIPS 2024), and novel diffusion model guidance (ICLR 2025). This evolution demonstrates convergence of generative modeling, clinical data analysis, and protein design.

He actively mentors 24 PhD students across diverse AI domains:

• Current Research: Conversational agents, emotion analysis, clinical reasoning, antibody design, and diffusion model optimization
• Recent Graduates: Interpretable language models, biomedical semantics, task-oriented dialogue, and social media knowledge extraction

Research is supported by imec funding and collaborations with Flemish biotech companies, building on his post-doctoral experience securing media-sector projects. Within IDLab, he co-leads the Text-to-Knowledge cluster driving NLP innovations for healthcare, legal, and economic applications.