Detecting connectivity changes in neuronal networks

Survey and evaluation of neural computation models for bio-integrated systems

“…This review gives indications for selecting a method having features according to [27]. These features for a method to be: 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 -accurate independently from the heterogeneity in firing types of neurons in the network, -sensitive to fine grain changes, and -highly specific in order to decrease the number of false connections found.…”

“…The major issue about this class of methods is that they do not consider influences from other spike trains than the pair of trains being studied. Therefore, pair-wise estimates can show inaccurate results when considering highly interconnected networks as shown in [27,54]. For this reason, it is important to consider approaches that address the influences from the entire spike trains of the network when analyzing the connectivity within this network.…”

“…a spike train is considered to be modulated by others. This MRP is modeled in [54] and [27] as a hazard function expressing the conditional probability of a spike at time t relatively to all spiking activity of the network prior to that time. 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 Considering a network of size n+1, this proportional hazard function of spike train of neuron A, , is modulated in this model by the possible influence of other spike trains from other neurons B i of the network as: (10) In equation (10), represents the time spent from the last spike of neuron A.…”

“…Additionally, the few parameters from quadratic models are easily adjustable to fit with in-vitro recordings. Therefore, Izhikevich simple model of neuronal behavior [25,26] (see equation (3)) is currently favored by the research community as it shows very accurate dynamical behavior of neurons in comparison with reality [27]. While being accurate, the model of Izhikevich (equation (3)) is also computationally simple in comparison with the model from Hodgkin-Huxley.…”

Identifying Chaotic FitzHugh–Nagumo Neurons Using Compressive Sensing

“…In addition, the approach of random phase resetting was introduced to reconstruct the details of the network structure [12]. For neuronal systems, there was a statistical method to track the structural changes [13,14]. While many of these previous methods require complete or partial information about the dynamical equations of the isolated nodes and their coupling functions, completely data-driven and model-free methods exist.…”