Fitting of adaptive neuron model to electrophysiological recordings using particle swarm optimization algorithm

Abstract

In order to fit neural model’s spiking features to electrophysiological recordings, in this paper, a fitting framework based on particle swarm optimization (PSO) algorithm is proposed to estimate the model parameters in an augmented multi-timescale adaptive threshold (AugMAT) model. PSO algorithm is an advanced evolutionary calculation method based on iteration. Selecting a reasonable criterion function will ensure the effectiveness of PSO algorithm. In this work, firing rate information is used as the main spiking feature and the estimation error of firing rate is selected as the criterion for fitting. A series of simulations are presented to verify the performance of the framework. The first step is model validation; an artificial training data is introduced to test the fitting procedure. Then we talk about the suitable PSO parameters, which exhibit adequate compromise between speediness and accuracy. Lastly, this framework is used to fit the electrophysiological recordings, after three adjustment steps, the features of experimental data are translated into realistic spiking neuron model.

On the Use of a Multimodal Optimizer for Fitting Neuron Models. Application to the Cerebellar Granule Cell

“…An alternative to these methods are the EAs, such as GAs, and the PSO, which allow solving parameters tuning problems that classical methods might fail for multidimensional non-linear systems, such as the AdEx model. These algorithms provide high flexibility, universality (being able to be applied to different cases) and proved to be fast and efficient strategies to take into consideration for fitting neuron models (Cachón and Vázquez, 2015 ; Van Geit et al, 2016 ; Shan et al, 2017 ). This is the case of the optimization of an AdEx model of a cerebellar granule cell (GrC) and a Golgi cell (GoC) proposed in Nair et al ( 2015 ).…”

“…Large-scale neural network simulations composed of thousands or millions of neurons are useful for better understanding brain information processing primitives. Simplified single-neuron models of low computational cost and based on a few parameters have been proposed to reproduce neuronal firing patterns to encode and decode the information contained in electrophysiological recordings (Izhikevich, 2004 ; Shan et al, 2017 ; Marín et al, 2020 ). These models are required to meet efficiency and biological realism for hypothesizing the functional impact of relevant neuron properties within large-scale simulations.…”

Z-Shaped Transfer Functions for Binary Particle Swarm Optimization Algorithm

“…e particle swarm optimization (PSO) algorithm is one of the widely used evolutionary algorithms inspired by animal social behaviors [1,2]. It has the search speed, high efficiency, simple algorithm, and so on and has been widely used in crystal structure prediction [3], medical detection [4], grid scheduling [5], robot path planning [6], clustering problem [7], neural network, and many other areas [8][9][10]. However, many optimization problems have binary searching space, so it is necessary to develop binary optimization algorithm to solve them.…”