Intelligent behaviour fascinates me: how animals can efficiently process a flood of sensory signals and learn to survive. Nature has produced elegant solutions for information processing that I believe can help us tackle challenges in modern society. I’m passionate about taking what we know from how brains process information, to making intelligent machines.

I work in the Bayesian Intelligent Autonomous Systems lab of the TU Eindhoven. We design agents that learn to solve tasks by interacting with their environment. Our design philosophy is based off of the Free Energy Principle, a leading theory of how the brain processes information. Our agents are deployed to signal processing tasks, where they filter noisy sensor data, and to control tasks, where they plan sets of actions to achieve a specified goal. My focus is on bringing our agents to mobile robotics.

Previously, I worked on the theoretical limitations of machine learning, in particular sampling bias. I tried to understand how, when and why algorithms fail to generalize from a training sample to real-world settings. I designed robust estimators, which have been applied to image, signal and natural language processing problems.