prof. dr. ir. J.H.P. Kwisthout (Johan)

I am internationally recognized as one of the experts on computational complexity aspects of Bayesian networks, which was also the topic of my PhD research. After my PhD period I specialized in the complexity of approximating the most probable explanation for observed variables in such networks. Currently, my focus is on PGMs for Decision Support Systems and on Foundations of Stochastic Computing. The first line will focus on explainability, trustworthiness, maintainability, online or federated learning etc. in Bayesian networks and other PGM models, particularly with an application in clinical decision support systems. Key project in this line is in the Personalised Care in Oncology consortium of which I am programme leader. Furthermore, I am work package co-lead in the Healthy Data initiative. The second research line will focus on topics such as approximate Bayesian inference, (parameterized) complexity classes for stochastic computing, and realization of probability distributions and computations on them in novel materials and computing architectures. More particularly, we study the parameterized tractability of approximate Bayesian inferences, we search for novel approximation techniques, and the study how to formalize the extension of PGMs with new variables or values of variables. A novel interest is in implementing resource-bounded stochastic computations in novel materials and in brain-inspired neuromorphic systems. Our fundamental work in this area is in particular in formalizing a novel computational complexity theory for neuromorphic hardware, designing neuromorphic algorithms, and implementing Bayesian inferences in neuromorphic hardware. A new project in this line is in designing new MCMC algorithms using covarying random bits using magnetic fields.

• Natural Computing & Neurotechnology
• Perception, Action and Decision-making

• Probabilistic Graphical Models