I am interested in foundations of neural networks and models of computation in the brain. Deep neural networks have now established themselves as a powerful method in machine learning. However, the theoretical foundations of neural networks are poorly understood. My research is concerned with understanding how parameters of deep networks (such as depth, number of neurons, connectivity, etc..) influence the performance of the network and to use this information to derive novel training algorithms.
My other research interest is stochastic modelling. I am particularly interested in biological systems as natural information processors. The ability of biochemical networks to compute seems to be limited by the stochastic fluctuations these systems. Overcoming fluctuation usually comes at the cost of higher energy dissipation. For some biological systems it is also possible to define a "speed" of computation. It is then possible to formulate a speed-cost-accuracy trade-off of the "biological computer."
I am also interested in the limits to using science to solve societal problems. In the public (and indeed scientific) discourse the "scientific method" is often equated with "rationality." I believe that this view is naive and have expressed my thoughts on this in my popular science book The Science Myth: God, society, the self and what we will never know .
My research publications can be found here. Beside my research papers, I published together with David Barnes a text book Guide to Simulation and Modeling for Biosciences to help students make their first steps in biological modelling.
I have an emergent interest in deep neural networks and deep learning. If you would like to do a PhD in this area, please contact me to discuss possible topics. There is funding available for students from the UK or the European Union.
01/01/2017