Title: A model of the starburst amacrine cell for motion direction detection

Authors: Fenggang Yuan; Hiroyoshi Todo; Cheng Tang; Zheng Tang; Yuki Todo

Addresses: Faculty of Engineering, University of Toyama, Toyama, 930-8555, Japan ' Wicresoft Co., Ltd, Tokyo, Japan ' Department of Electrical and Mechanical Engineering, Nagoya Institute of Technology, Nagoya, 466-8555, Japan ' Faculty of Engineering, University of Toyama, Toyama, 930-8555, Japan ' Faculty of Electrical, Information and Communication Engineering, Kanazawa University, Ishikawa, 920-1192, Japan

Abstract: The mechanism of motion direction detection for direction selective ganglion cells (DSGCs) is still not well-understood and under debate. Recent studies have elaborated the critical experimental evidence that the starburst amacrine cells (SACs) can trigger off the null-direction inhibition to DSGCs. In this study, a simple but effective neural model is introduced for the SACs to solve the motion direction detection problems, based on greyscale images in the visual scene. Virtual simulations demonstrate that the neural model is capable of detecting the motion direction of objects with different shapes, sizes, greyscales, and positions efficiently. To further demonstrate the feasibility and effectiveness of the model, the performance of the proposed model is compared with traditional artificial neural networks (ANNs). Experimental results show it can completely beat ANNs on motion direction detection problems, in terms of recognition accuracy, noise immunity, computational and learning costs, biological soundness, and reasonability.

Keywords: deep learning; convolutional neural network; CNN; perceptron; direction detection; greyscale; artificial neural network; ANN.

DOI: 10.1504/IJBIC.2023.130560

International Journal of Bio-Inspired Computation, 2023 Vol.21 No.2, pp.69 - 80

Received: 14 Nov 2022
Accepted: 11 Dec 2022
Published online: 27 Apr 2023 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article