Large scale parameter estimation for nonlinear dynamic systems: Application on spike-in, spike-out neural models

This paper presents a general method of parameter estimation for large-scale non-linear dynamic models a with particular focus on parameter estimation for spike-in, spike-out neural models. The aim is to provide a convex optimization algorithm for tuning parameters of such a model which enables solving large-scale estimation problem in a linear time. Parameter estimation for a single layer neural network containing hundreds of synapses is addressed and efficiency/performance of the proposed methodology is demonstrated by solving a few examples. It will be also demonstrated that parameters of the model for mapping CA3 output of hippocampus cell into CA1 output, under patch clamp experiment, can be successfully estimated by utilizing the methodology of this paper.