SCL Seminar by Igor Franovic
SCL seminar of the Center for the Study of Complex Systems, will be held on Thursday, 24 May 2018 at 14:00 in the library reading room “Dr. Dragan Popović" of the Institute of Physics Belgrade. The talk entitled
"Switching dynamics in networks of stochastic rate-based neurons"
will be given by Dr. Igor Franović (Scientific Computing Laboratory, Center for the Study of Complex Systems, Institute of Physics Belgrade).
Abstract of the talk:
Macroscopic variability is an emergent property of neural networks, typically reflected in the onset of slow rate oscillations, which comprise spontaneous switching between the episodes of elevated neuronal activity, called UP states, and the quiescent episodes, referred to as DOWN states. Such dynamics is fundamental to activity of cortical networks because it facilitates or mediates different forms of learning and memory. In this talk, we will present recent results on paradigmatic scenarios for switching dynamics, discuss their dependence on the ingredients inherent to neuronal systems, including noise, coupling delays and heterogeneity in network topology. Collective dynamics of clustered networks of stochastic rate-based neurons will be considered within the framework of an effective model, describing macroscopic behavior by coupled mean-field models representing each of the clusters [1,2]. We will illustrate how bifurcation analysis of the effective model in the thermodynamic limit provides insight into the structure and transitions between the metastable states shaping the spontaneous and the induced network dynamics. In particular, it will be demonstrated how one can estimate the different contributions to effective macroscopic noise [1,2], explain the mechanisms behind switching dynamics and predict the relevant parameter domains [1,3], as well as understand the impact of clustering on the spontaneous [1] and induced network dynamics [4]. As opposed to statistically homogeneous random networks, where in the thermodynamic limit one finds only monostable or bistable behavior [2,3], we will show that clustering promotes multistability, thereby enhancing the robustness of the switching phenomenon [1].
[1] I. Franović and V. Klinshov, Chaos 28, 023111 (2018).
[2] V. Klinshov and I. Franović, Phys. Rev. E 92, 062813 (2015).
[3] I. Franović and V. Klinshov, EPL 116, 48002 (2016).
[4] I. Franović and V. Klinshov: “Stimulus-evoked activity in clustered networks of stochastic rate-based neurons“, submitted (2018).
"Switching dynamics in networks of stochastic rate-based neurons"
will be given by Dr. Igor Franović (Scientific Computing Laboratory, Center for the Study of Complex Systems, Institute of Physics Belgrade).
Abstract of the talk:
Macroscopic variability is an emergent property of neural networks, typically reflected in the onset of slow rate oscillations, which comprise spontaneous switching between the episodes of elevated neuronal activity, called UP states, and the quiescent episodes, referred to as DOWN states. Such dynamics is fundamental to activity of cortical networks because it facilitates or mediates different forms of learning and memory. In this talk, we will present recent results on paradigmatic scenarios for switching dynamics, discuss their dependence on the ingredients inherent to neuronal systems, including noise, coupling delays and heterogeneity in network topology. Collective dynamics of clustered networks of stochastic rate-based neurons will be considered within the framework of an effective model, describing macroscopic behavior by coupled mean-field models representing each of the clusters [1,2]. We will illustrate how bifurcation analysis of the effective model in the thermodynamic limit provides insight into the structure and transitions between the metastable states shaping the spontaneous and the induced network dynamics. In particular, it will be demonstrated how one can estimate the different contributions to effective macroscopic noise [1,2], explain the mechanisms behind switching dynamics and predict the relevant parameter domains [1,3], as well as understand the impact of clustering on the spontaneous [1] and induced network dynamics [4]. As opposed to statistically homogeneous random networks, where in the thermodynamic limit one finds only monostable or bistable behavior [2,3], we will show that clustering promotes multistability, thereby enhancing the robustness of the switching phenomenon [1].
[1] I. Franović and V. Klinshov, Chaos 28, 023111 (2018).
[2] V. Klinshov and I. Franović, Phys. Rev. E 92, 062813 (2015).
[3] I. Franović and V. Klinshov, EPL 116, 48002 (2016).
[4] I. Franović and V. Klinshov: “Stimulus-evoked activity in clustered networks of stochastic rate-based neurons“, submitted (2018).
