BIOINSPIRED APPROACH TO MODELING RETINAL GANGLION CELLS USING SYSTEM IDENTIFICATION TECHNIQUES

Abstract

The processing capabilities of biological vision systems are still vastly superior to artificial vision, even though this has been an active area of research for over half a century. Current artificial vision techniques integrate many insights from biology yet they remain far-off the capabilities of animals and humans in terms of speed, power, and performance. A key aspect to modeling the human visual system is the ability to accurately model the behavior and computation within the retina. In particular, we focus on modeling the retinal ganglion cells (RGCs) as they convey the accumulated data of real world images as action potentials onto the visual cortex via the optic nerve. Computational models that approximate the processing that occurs within RGCs can be derived by quantitatively fitting the sets of physiological data using an input–output analysis where the input is a known stimulus and the output is neuronal recordings. Currently, these input–output responses are modeled using computational combinations of linear and nonlinear models that are generally complex and lack any relevance to the underlying biophysics. In this paper, we illustrate how system identification techniques, which take inspiration from biological systems, can accurately model retinal ganglion cell behavior, and are a viable alternative to traditional linear–nonlinear approaches.

Existing System

Nonlinear autoregressive network with exogenous inputs (NARX) and k-nearest neighbors (kNNs) approaches.

Proposed System

In the experiments here, we expand on by introducing, in addition to the NARMAX model, the self-organizing fuzzy neural network (SOFNN) and NARX methodologies. The predictive performance of the investigated methodologies to adequately model a retinal ganglion cell’s output is evaluated. Deriving a quantitative relationship between stimulus and response of an RGC is challenging if we consider the internal cell structure that precedes them or the numerous interactions over the many interconnections between cells. To simplify this, we consider the problem with a black-box approach, which aims to estimate a mathematical model for a regression data set and apply a number of different methods to form this model. In keeping with traditional approaches, the LN cascade approach is also utilized as a comparison to the investigated approaches.

Architecture

Layer structure of SOFNN

Internal structure of EBF neuron