视觉运动感知的神经网络研究

本文涉及到视觉图象信息加工与感知过程中一个十困难的问题:视觉运动信息的神经计算。文中介绍了蝇视觉系统的图形-背景相对运动分辨网络的基本框架,集中描述了图形-背景分辨系统的空间神经生理整合作用的计算机实验结果,证明在图形-背景分辨系统中的增益控制机制、突触传递的非线性、双眼之间的相互作用,以及池细胞的方向选择性等,在空间生理整合中起着重要作用。 文中以Reichardt的相关型初级运动检测器二维阵列和Kohonen的自组织特征映射网络相结合,构成了一个视觉运动感知的自组织神经网络。计算机仿真表明,通过无监督学习,网络可以学会解决局域运动检测所带来的多义性问题。并在网络的输出层自组织生成了类似灵长类视皮层MT区的运动功能柱结构。结果表明:时空互相关、映射、合作、竞争、自适应等可能是运动感知网络的几个基本的神经原理。     

生命与思维——在混沌的边缘演化

“道可道，非常道；名可名，非常名。无，名天地之始；有，名万物之母”。（老子，道德经）可以说，现代复杂性科学，实际上是在探索中国古代“道德经”中所称的“众妙之门”。在复杂性科学中，“混浊的边缘”和“自组织临界性”是两个很深刻的概念，此两者亦可谓“同出而异名”（道德经）。它们是近一二十年来，以美国桑塔费研究所的学者们为代表的科学思想的结晶，是对线性思维和还原论的反潮流，它引导我们换一个方式并用一个新的视角，去观察与思考整个自然界和人类社会，乃至去观察产生了生命和自我意识的整个宇宙。一、“混饨的边缘”…     

神经网络研究和未来的科学革命

神经网络研究可以追朔到四十年代信息科学的开创时期,它在走过了半个世纪的曲折发展道路之后,正迎来一个空前活跃的新的发展时期。神经网络研究已成为现代神经科学、数理科学及信息科学等综合研究领域中共同的科学前沿之一。这不仅体现了近二三十年来自然科学发展的重要特征,也体现了科学演化的必然趋势-研究非线性、探索复杂性。神经网络系统是由大量的形式神经元联结而成的高度错综复杂的非线性系统。结点特性、联结拓扑结构及学习规律是确定一个神经网络的三要素。神经网络在功能上反映了生物神经系统的若干基本特性,但并不是脑的逼真描写,它只是脑功能的某种简化、抽象和模拟。神经网络研究的目标是阐明脑的工作原理和研制具有某些大脑神经计算功能的人工智能系统。这     

Neural Network Approaches to Visual Motion Perception

This paper concerns certain difficult problems in image processing and perception: neuro-computation of visual motion information. The first part of this paper deals with the spatial physiological integration by the figure-ground discrimination neural network in the visual system of the fly. We have outlined the fundamental organization and algorithms of this neural network, and mainly concentrated on the results of computer simulations of spatial physiological integration. It has been shown that the gain control mechanism , the nonlinearity of synaptic transmission characteristic , the interaction between the two eyes , and the directional selectivity of the pool cells play decisive roles in the spatial physiological integration. In the second part, we have presented a self-organizing neural network for the perception of visual motion by using a retinotopic array of Reichardt's motion detectors and Kohonen's self-organizing maps. It .has been demonstrated by computer simulations that the network is abl     

IDENTIFICATION OF DYNAMIC BEHAVIOUR OF LIGHT-POTENTIAL TRANSDUCTION SYSTEM IN PHOTORECEPTOR OF CALLIPHORA ERYTHROCEPHALA USING IRS-STIMULI

The linear dynamic behaviour of the light-potential transduction system in photorceeptor of Calliphora erythrocephala has been studied through the cross-correlation method using InverseRepeat Sequences and the lineal model of phototransduction is provided. The impulse response function and frequency characteristics have been determiued in the condition of background illumination.The results suggest that the R1-6 is equipped with a gain and time constant control mechanism which depends upon the adaptation level of the system under study.It becomes evident that the IRS correlation method is powerful to quantify the dynamic characteristics of the system containing nonlinearities and will become a very popular application for the study of biological sensory systems.The visual receptor cells R1-6 in the fly are the most sensible choice for the study of the light-potential transdnction system of the eye. The input of the system is the temporarily varying light intensity applied upon the eyes of the fly and th     

NEUROCOMPUTATION OF VISUAL RELATIVE MOTION INFORMATION

In this paper the neurocomputation of figure-ground relative motion information in thevisual system of the fly have been investigated in great detail by a combination of quanti-tative behavioural experiments and computational model simulations. Only torque responsesabout the vertical axis of the tethered flying flies (Musca domestica) were determined inthe behavioural experiment. The main results of behavioural experiments are: (i) The dynam-ics of the torque responses depends not only on the phase relationship between figure andbackground motion but also on the oscillation frequency of the figure and ground. (ii) Inall the phase relations tested, the time courses are a characteristic fingerprint of the partic-ular phase relationship. (iii) The variation of the amplitude of the response peaks is an espe-cially sensitive indicator for the variability of figure--ground discrimination behaviour. The main results of computer simulations are: (i) The "computer fly", the networkmodel of both the SF-system and     

视觉相对运动的神经计算原理

本文将定量行为实验与计算机仿真相结合,研究了蝇视觉系统的图形与背景相对运动信息的神经计算原理。在定量行为实验中仅检测了作定点飞行的蝇绕其垂直轴的扭矩信号。行为实验的主要结果是:(1)扭矩反应的动态特性不仅与图形和背景运动之间的相位有关,也与振动频率有关;(2)在全部检测的相对相位中,反应时程是给定的相位关系的“指纹”特征;(3)主尖峰反应幅度的变化是图形-背景分辨行为的多样性的灵敏的指标。计算机仿真的主要结果为:(1)“计算机蝇”,可描述至今为止在不同的相对运动条件下的扭矩反应的主要特征;(2)在不同刺激条件下的计算机仿真与相应的行为实验在其“指纹”特征方面完全符合;(3)计算机仿真中的反应时程与参数n,q,T和k密切相关。 本文的行为实验与计算机仿真在神经网络层次上加深了对视觉相对运动信息加工策略的理解。     

COMPUTER SIMULATIONS OF FIGURE-GROUND DISCRIMINATION IN THE VISUAL SYSTEM OF THE FLY

Figure-ground discrimination and pattern discrimination represent one of the most important problems in computer vision. Based on computational neuronal network model proposed by Reichardt et al. for figure-ground discrimination in the visual system of the fly, entire systematic computer simulations were carried out under nonstationary conditions. Results show that the model network and simulations have predictive power for behavioural experimental results. This paper proposes that motion perception and elementary pattern discrimination in human visual system may be mediated by some kind of figure-ground system with movement detectors as input layer.     

物理科学与脑科学的联姻——脑与精神奥秘的新探索

揭示脑的奥秘,已经不再是乌托邦式的空想了;复杂系统的自发的集体行为,似乎生来就是难以预测的;脑是怎样应付、适应和表达多样的,多变的外部世界的;选择性学习可以看作是对能量地形的树结构修枝的过程;混沌可能是大脑不同于现有人工智能系统的主要特征;实现对自然和人类自身认识的新飞跃,实现从必然王国向自由王国的新飞跃;迎接新的更伟大的科学革命——智能革命的到来.