Automatic Tuning of a Retina Model for a Cortical Visual Neuroprosthesis Using a Multi-Objective Optimization Genetic Algorithm

Abstract

The retina is a very complex neural structure, which contains many different types of neurons interconnected with great precision, enabling sophisticated conditioning and coding of the visual information before it is passed via the optic nerve to higher visual centers. The encoding of visual information is one of the basic questions in visual and computational neuroscience and is also of seminal importance in the field of visual prostheses. In this framework, it is essential to have artificial retina systems able to function in a way as similar as possible to the biological retinas. This paper proposes an automatic evolutionary multi-objective strategy based on the NSGA-II algorithm for tuning retina models. Four metrics were adopted for guiding the algorithm in the search of those parameters that best approximate a synthetic retinal model output with real electrophysiological recordings. Results show that this procedure performs well when different trade-offs has to be considered during the design of customized neuro prostheses.

Control methodologies for vibration control of smart civil and mechanical structures

“…The GA is often preferred over gradient‐based optimization methods because it uses the crossover and mutation operations to search in multiple directions thus avoiding entrapment in a local optimum (Rostami & Neri, ; Wright & Jordanov, ). More recently, many articles have been published on multiobjective GA (Carrillo, Jiang, Rojas, & Valenzuela, ; Martinez‐alvarez et al, ; Rostami, Neri, & Epitropakis, ; Wang, Liu, Yuan, & Chen, ; Yang, Emmerich, Baeck, & Kok, ) and many‐objective GA (Pan, He, Tian, Su, & Zhang, ).…”

“…The GA is often preferred over gradient-based optimization methods because it uses the crossover and mutation operations to search in multiple directions thus avoiding entrapment in a local optimum (Rostami & Neri, 2016;Wright & Jordanov, 2017). More recently, many articles have been published on multiobjective GA (Carrillo, Jiang, Rojas, & Valenzuela, 2018;Martinez-alvarez et al, 2016;Rostami, Neri, & Epitropakis, 2017;Wang, Liu, Yuan, & Chen, 2018;Yang, Emmerich, Baeck, & Kok, 2016) and many-objective GA (Pan, He, Tian, Su, & Zhang, 2017). Kundu and Kawata (1996) present an optimal feedback controller design based on the GA, in which the performance function of a control system in Equation (3) is decomposed as a multiple-criteria optimization problem thus avoiding the weigh matrices required in the traditional LQR control.…”

Metaheuristic Optimisation Algorithms for Tuning a Bioinspired Retinal Model

“…Synthetic signals have to be unequivocal and speedy to reproduce the activity of a restricted number of retinal ganglion cells (RGCs), which send processed visual information through the optic nerve to higher visual centres. Considering those points, we have extended a highly parametrised bioinspired retinal model framework, first presented in [6], to explore new tuning possibilities. This bioinspired mathematical model describes the different stages that comprise the vertebrate retina, simplifying its behaviour to generate a low latency model with high throughput in real-time.…”

Parameter Optimization Using Covariance Matrix Adaptation—Evolutionary Strategy (CMA-ES), an Approach to Investigate Differences in Channel Properties Between Neuron Subtypes

“…Most of the modern search methods use variants of evolutionary algorithms (Vanier and Bower, 1999 ; Keren et al, 2005 ; Hendrickson et al, 2011b ; Brookings et al, 2014 ; Martínez-Álvarez et al, 2016 ; Rumbell et al, 2016 ; Martínez-Cañada et al, 2017 ; Neymotin et al, 2017 ). The covariance matrix adaptation evolutionary strategy is a modern evolutionary algorithm that works quite well for large numbers of parameters (Hansen and Kern, 2004 ).…”