Biography
Wouter Kouw develops probabilistic machine learning systems, particularly active inference agents, for autonomous robots. By formulating Bayesian inference as message passing on graphical models, he builds hierarchical reactive agents that learn internal kinematics and adapt to uncertain dynamic environments. These agents are deployed on mobile robots and high-tech systems to automate hazardous tasks, creating safer work environments for people.
“Intelligence is more than just computation at scale - it’s about making sense of the world under uncertainty.”
- Autonomous intelligent systems
- Probabilistic machine learning
- Neuromorphic computing
- Biomimetic robotics
-
PhD in Computer Science, 2018
Delft University of Technology
-
MSc in Neuroscience, 2013
Maastricht University
Projects
Publications
Gaussian variational inference with non-Gaussian factors for state estimation.
IEEE Robotics & Automation Letters, 2026.
Effects of priors on epistemic uncertainty in autoregressive active inference.
EurIPS Workshop on Epistemic Intelligence in Machine Learning, 2025.
Bayesian autoregression to optimize temporal Matérn-kernel Gaussian process hyperparameters.
International Conference on Probabilistic Numerics, 2025.
Message passing-based inference in an autoregressive active inference agent.
International Workshop on Active Inference, 2025.
Software
RxInfer is a powerful Julia package for event-driven variational Bayesian inference, and is used for probabilistic machine learning, signal processing, adaptive control and the design of intelligent agents.