Max Planck Institute for Dynamics and Self-Organization -- Department for Nonlinear Dynamics and Network Dynamics Group
Personal tools
Log in


Tuesday, 14.07.2015 17 c.t.

Spatio-temporal processing in biological and artificial recurrent networks

by Dr. Andreea Lazar
from Ernst Strüngmann Institute (ESI) for Neuroscience in Cooperation with Max Planck Society, Frankfurt/M.

Contact person: Viola Priesemann


MPI DS seminar room (0.77/0.79)


Recurrent circuits represent time implicitly through the effect that is has on processing. When a stimulus is presented, the evolution of the network activity can be described as a neural trajectory through the network’s state space, beginning and ending at a point that represents baseline activity. Because of this memory property, recurrent neural networks can exploit long-term dependencies in the input when calculating the output and perform well on complex temporal tasks, for which feed-forward architectures are ill-equipped. However, training or optimizing such models can be challenging. It has been proposed that recurrent networks whose dynamics are situated at the edge of chaos demonstrate improved computational capabilities. Interestingly, cortical networks have been shown to exhibit similar critical dynamics. In contrast to most of their artificial counterparts, biological networks are continuously refined by various forms of plasticity that act in a reproducible, highly specific fashion. We explore the interplay between structure, dynamics and computational power in brain-inspired self-organizing recurrent networks (SORNs) shaped by plasticity. Additionally, we analyze the spatio-temporal properties of neuronal activity recorded from the primary visual cortex and explore the signatures of recurrent computation via methods suitable for the analysis of high dimensional dynamics.

back to overview