鸽视顶盖快速显著感知编码模型研究  

Neural Coding Model for Fast and Significant Perceptual in the Pigeon Optic Tectum

作　　者：王松伟[1] 黄淑漫 师丽[1,3] 王梦珂[2] Wang Songwei;Huang Shuman;Shi Li;Wang Mengke(School of Electrical Engineering,Zhengzhou University,Zhengzhou 450001,China;Department of Automation,Tsinghua University,Beijing 100084,China;Industrial Technology Research Institute,Zhengzhou University,Zhengzhou 450001,China)

机构地区：[1]郑州大学电气工程学院,河南郑州450001 [2]郑州大学产业技术研究院,河南郑州450001 [3]清华大学自动化系,北京100084

出　　处：《系统仿真学报》2018年第11期4086-4099,共14页Journal of System Simulation

基　　金：国家自然科学基金(U1304602)

摘　　要：在经典目标识别理论中,哺乳动物的视网膜和LGN中许多神经元执行了DoG (Difference of Gaussian)操作,其功能一般认为是白化,祛除冗余,并增强了边缘。通过对鸽视顶盖的ON-OFF神经元进行电生理研究,发现其利用FSL(First-SpikeLatency)进行场景整体信息粗略快速的传递,利用发放率对场景中显著性特征进行相续的传递。通过解析神经元响应模式,提出了OT(Optic Tectum)的ON-OFF神经元工作机制的一种假设,搭建了模型架构。该研究对新型的Spike神经网络的研究具有一定的借鉴意义。In the classical target recognition theory, many neurons in the mammalian retina and LGN perform DoG operations. Their functions are generally considered to be whitening, eliminating redundancy, and enhancing edges. Where is the redundant information？ In this paper, the electrophysiological study of the ON-OFF neurons of the dove optic roof was performed. It was found that the FSL was used to carry out the rough and fast transfer of the whole scene information, and then the salient features in the scene were successively transmitted using the issue rate. For this reason, by analyzing the response pattern of neurons, this paper proposes an assumption of the ON-OFF neuron working mechanism of OT （Optic Tectum） and builds a model architecture. This research has certain reference significance for the research of the new Spike neural network.

关 键 词：视顶盖 显著性表征 spike神经网络 延时编码 

分 类 号：Q424[生物学—神经生物学]