Homepage


M. Hasler

Swiss Federal Institute of Technology Lausanne (EPFL), Lausanne, Switzerland

Dynamik in Netzwerken -
Gedanken zur Informationsverarbeitung
in biologischen und technischen Netzwerken

Veranstaltungsreihe: Orte der Dynamik

Time: April 20, 2004, 5:00 pm
Place: Building 9, Kleiner Physik-Hörsaal


Anhand der Aufarbeitung von vielen, gleichzeitig gemessenen Signalen aus dem Gehirn eines Affen, welche fuer die Steuerung eines Robotersarms verwendet werden, kann man schliessen, dass Information im Gehirn auf viele Neuronen diffus verstreut und vermischt ist. In Analogie dazu kann man neuere Entwicklungen in der Informatik sehen, welche verteiltes, ueber das Internet verstreutes Rechnen und Informationsaustausch verfolgen. Einige prinzipielle Ueberlegungen ueber moegliche Informationsverarbeitung mittels Netzwerkdynamik werden angestellt.



Titel: Dynamics in networks: thoughts about information processing in biological and technical networks.

Abstract:
It appears that information items in the brain are distributed over many neurons, and most neurons process a mixture of such items. Consequently, in order to extract specific information, one has to process simultaneously signals from many neurons. This is illustrated on the example of motor commands extracted from the cortex of a Macaque monkey and used to control a robot arm (experiments performed at the laboratory of Miguel Nicolelis at Duke University). While these ground-breaking experiments have clear practical implications, much more understanding about the nature of information processing in the brain is needed in order to progress substantially in the domain of brain-machine interfaces.

Distributed information processing is also a modern paradigm in computer science. First practical realizations, using the Internet, are peer-to-peer file exchange systems such as Napster and Kazaa, or Grid computing, where local computing resources are coordinated over the network to solve a complex task. On a more fundamental level, we shall show a few basic mechanisms of nonlinear dynamics in networks that could be used for information processing.

Maybe the future will bring much more sophisticated distributed computing on top of large heterogeneous and time varying networks composed of unreliable elements such as the Internet. We think that in order to go in this direction the field of distributed computing should get some inspiration from neuroscience just as this was the case of artificial neural networks towards the end of last century. Vice versa, distributed computing, already at the service of neuroscience through Grid computing, could also be a source of inspiration for neuroscience.

Homepage