Christian Mayr

Postdoc -- ended Apr 2015
Work phone:
+41 44 63 53066
The brain constantly receives a large input data stream from its environment. It manages to correctly sort, process and react to all stimuli impinging on it. This ability is evident at all levels of neuronal organization, down to cell level, where a single neuron receives input from up to 10000 upstream neurons. As a neuroscientist, I am fascinated by this parallel, error-tolerant, self-organizing information processing capability. As an engineer, I aim to adapt this processing to neurally-inspired, brain-like computing (so-called neuromorphic circuits), bringing new computational paradigms to conventional CMOS circuit design.
To study neuronal information processing, I am designing hybrid couplings of neuron cell cultures with neuromorphic integrated circuits. These hybrid setups give significantly more detailed control of e.g. topology and dynamics than an in-vivo setup, allowing in-depth study of various aspects of neural tissue behaviour.
In terms of neuromorphic circuit design, I am especially interested in nanoscale circuits and devices with their corresponding high computational density. At this density (e.g. >1e9 transistors on a standard processor in a 28nm CMOS technology), it becomes increasingly difficult to implement conventional serial processing paradigms. Biology in this respect points the way towards parallel paradigms that are insensitive to device variations and encompass various principles of self-construction and organization. This will allow us to harness the large computational potential offered by nanoscale CMOS and novel nanoscal devices.










© 2017 Institut für Neuroinformatik