BayesBrain, an Exploratory Multidisciplinary AI Research (EMDAIR) project, funded by the Eindhoven Artificial Intelligence Systems Institute

With our brains we perform parallel computer tasks, such as adapting and learning. Could it be that similar processes also take place in computers? Much research is already being done into new brain-inspired computation in the development of so-called neuromorphics chips.

Regina Luttge & Bert de Vries, TU Eindhoven / BayesBrain

BayesBrain, led by TU/e researchers Regina Luttge and Bert de Vries, is trying a slightly different approach, because we still know far too little about how our brains actually work. ‘We therefore want to use existing computer architecture to control the brain and make it understandable. Of course, this cannot be done directly on a human brain. At the same time, we do not want to do our research with laboratory animals – even though that might be another approach – we want to do it with a well-defined system of a few human brain cells, i.e., Brain-on-Chip, like we also developed in the framework of hDMT for learning about biology and medicine.

In BayesBrain, we will explore the possibility of growing stem cell-derived brain cells on a microfluidic chip that provides us with a hierarchical network structure. We will start with commercially available multielectrode arrays (MEAs) to monitor the electrical activity of these neuronal cultures and then also test whether we can write new functions into these networks. One such function, for example, is an instruction for controlling an inverted pendulum. The computer can already calculate this classical control problem, but it is terribly inefficient at it; the human brain can do this (and at the same time tell a story on a stage, for example see, a juggler in the theatre)! The brain can do this because it has evolved according to the principle of free energy minimization. This project will replace part of the existing computational architecture and break into the neuronal culture and then send the response of the cells back to the computer for a new prediction, and if the pendulum still remains upright, we will celebrate a new era in computing and push the boundaries Artificial Intelligence. Which in turn, can enable better medicine on a long run’.

For more information click here.

 

 

Connect with us