Centre for Complex Systems Events

The Centre for Complex Systems participates in, sponsors and hosts a number of events throughout the year.

Two of the Centre's major events programs are the C³ Symposia, and the C³ Research Camps.

Reconstructing neural networks and classifying synapses from spike train data using local transfer entropy

Date: Friday 23 November 2018
Time: 11am
Venue: School of IT Building J12, Lecture Theatre Rm 123 (Ground floor via Wintergarden)
Presenter: Felix Goetze, National Central University, Taiwan

Abstract: Transfer entropy has become a standard technique for inferring the effective connectivity in complex systems. Being a model-free method and therefore capturing both the linear and non-linear directed interactions between variables, it is useful for a wide variety of systems. Applied to spike train recordings from neurons this technique allows to reconstruct the neural network, since significant information transfer between two neurons is indicative of an underlying synapse, which could be be either excitatory or inhibitory (Nigam et al, The Journal of Neuroscience 2016). We extend on this technique by distinguishing these interaction types by analyzing the local transfer entropies (Lizier et al, Physical Review E 2008) which can quantify whether a spike in a presynaptic neuron is informative or misinformative about a postsynaptic spike. Furthermore we discuss the use of state-conditioned transfer entropy estimates (Stetter et al, PLOS Computational Biology 2012) from neural dynamics with intermittent bursting activity to reduce spuriously measured interactions during these highly synchronized events. The improved performance of these network reconstructions and synapse classifications are tested on simulated spike trains of randomly connected neurons with random interaction delays using the Izhikevich model, similar to a previous study (Ito et al, PLOS One 2011).

Bio: Felix Goetze is a PhD candidate in physics at National Central University and in the Taiwan International Graduate Program at Academia Sinica in Taipei. His research is on the quantification of information transfer among neurons to reconstruct neural networks from their spike trains. He completed a Master's degree at National Central University under the Taiwan scholarship doing research on the nonlinear dynamics of neural networks. For his Bachelor's degree he studied at Technical University of Darmstadt in Germany, for one year at the Hong Kong University of Science and Technology and did research at the Paul Ehrlich Institute on computational epidemiology.