Previous 
Up 
Next
Realistic modeling of neural networks typically involves simulating neurons and synaptic transmission at a significantly finer level of detail, compared to the very simple components of ANN:s. The connectivity is also more elaborate than in the artificial models. In this manner it becomes more straight-forward to incorporate neurobiological knowledge in our simulated networks. The simulation output can then be compared to the results from experimental recordings. Modeling of large networks of this sort has been made possible only recently, thanks to the high computational power of today's computers. Models of biological systems can also be of an intermediate complexity, between that of realistic models and artificial ANN:s. Such models play an important role in bridging the gap between theory and reality and pinpoint the critical mechanisms underlying a certain function.
Our main efforts along these lines are centered around three projects. Two are in collaboration with the Nobel Institute for Neurophysiology at Karolinska Institutet (Grillner, Wallén, Brodin). The first of these aims at the development of high quality software for realistic neural modeling, while the second deals with designing and simulating models of locomotion generation and control. The third major project involves investigation into cortical associative memory.
Next