Study of mesoscopic descriptions of neural networks

Abstract

Starting from the microscopic description of neural networks, we will derive mesoscopic models for non-stationary inputs. Our aim is to obtain numerical and/or analytical estimates of the parameters that describe the mesoscopic dynamics (neural mass models) in terms of the microscopic model parameters. Once this is obtained we will study how do the mesoscopic parameters change when the inputs which feed the network are non-stationary. During the first weeks, the student will create numerical codes to model the neural network and will learn analytical techniques to deriv

The point of the project is to know at which conditions the neural mass model corresponds to the neural network. At variable input, non-zero, the neural mass model does not correspond to the neural network

Se trata de ver en qué condicones se corresponden modelo de masa neuronal y modelo de red neuronal.

Es tracta de comprovar en quines condicions el model de massa neural es correspon amb el model de xarxa neuronal. Quan es rep un input variable extern diferent de zero, el sistema no és markovia, de mode que té memòrial. Així, les simulacions no surten iguals per input creixent pel neural mass model i el model de xarxa neural.