Bert de Vries

Bert de Vries

Professor

BIASlab, TU Eindhoven

Bert de Vries is Professor of Natural Artificial Intelligence at Eindhoven University of Technology (TU/e) and Principal Scientist at GN Hearing. His work is driven by a central question: how can we build artificial systems that operate robustly in uncertain, real-world environments? He approaches this challenge through Bayesian principles, and in particular the Free Energy Principle, which he develops into a practical framework for engineering scalable, adaptive AI systems.

At TU/e, he leads the BIASlab research group, where the focus is on translating theoretical insights from Bayesian brain theory into working computational architectures. A key outcome of this effort is the open-source toolbox RxInfer, which enables real-time, distributed Bayesian inference through reactive message passing on factor graphs. This work aims to establish a new computational paradigm for physical AI, in which perception, learning, and control are unified within a single inference process. A major application domain is the development of Bayesian hearing agents that continuously personalize context-aware hearing systems in situ.

De Vries received his MSc in Electrical Engineering from TU/e in 1986 and his PhD from the University of Florida in 1991. He began his career at SRI International and went on to hold research and leadership positions at Philips and GN Hearing. Since 2012, he has combined industrial innovation with academic research at TU/e, where he became full professor in 2021. In 2023, he co-founded Lazy Dynamics B.V., a deep-tech startup dedicated to bringing scalable Bayesian AI technologies into real-world applications.

Across academia and industry, De Vries has consistently worked at the intersection of theory and deployment. He holds 22 patents, has secured over €3 million in research funding, and has authored more than 100 scientific publications. His work contributes to a growing effort to reframe artificial intelligence as a problem of inference under uncertainty, and to develop the tools required to realize this vision in practice.

Interests
  • Bayesian Machine Learning
  • Signal Processing
  • Biomedical Applications
Education
  • PhD in Electrial Engineering, 1991

    University of Florida

  • MSc in Electrical Engineering, 1986

    Eindhoven University of Technology

Latest Publications

Active Projects