Analysis of a Network of Electrically Coupled Neurons Producing Rhythmic Activity in the Snail Helisoma Trivolvis

This paper deals with a quantitative analysis of a network of electrically coupled neurons (cyberchrons) in the snail Helisoma trivolvis. This network produces regular bursts of activity which drive the motoneurons controlling the feeding musculature. Analysis of the time constants and Bode plots obtained from current injection leads to the development of an equivalent model circuit describing the functional relationships of the individual neuronal elements. The equivalent model is used to demonstrate compensation of the frequency response of an individual neuron by the effective loading of the rest of the network. However, the current observed in a postsynaptic neuron is intergrated across the long cell time constant resulting in increased temporal and spatial summation. This frequency compensation and long duration of the postsynaptic response are both believed to be instrumental in the maintenance of a high frequency burst. This model will act as a framework on which active properties such as burst form, timing, and termination can be tested.     

Cerebellar Dynamics: The Mossy Fiber Input

A model for the cerebellar cortex is developed which qualitatively describes its temporal behavior. This neural network is treated as a lumped linear system with pure time delays between elements. The model predicts that the cerebellar cortex may act in part as a higher order lead-lag compensator. Physiological experiments performed in the vestibulo-cerebellum of the turtle appear to validate some of the conclusions of the model.     

Neuronal activities related to higher brain functions-theoreticaland experimental implications

The activities of several single units (6-10) were recorded simultaneously in the auditory cortex and in frontal cortical areas of cats and monkeys. The response properties of the single units and the interaction between them were studied. It is shown that single units in both areas may participate in prolonged processes and be involved in more than one process. Adjacent neurons need not function in unison; while some neurons are activated, others may stay inactive. The interactions among adjacent neurons are weak, and can be modulated by sensory stimulation, and by arousal and behavioral states. These properties lead us to hypothesize that information is represented in the cortex by coactivation of sets of neurons rather than by independent modulation of the single-unit firing rate. A single unit may be a member of several representing sets. Thus, each neuron may participate in more than one function and each small cortical area may contain members of several functional sets. A mechanism for computing and transmitting information, based on converging-diverging links, between neuronal sets is described and tested by simulations and analysis of experimental data.     

结构自适应自组织神经网络的研究

针对Ｋｏｈｏｎｅｎ的自组织特征映射（ＳＯＦＭ）神经网络的不足,本文把进化计算的思想用于神经网络的结构寻优之中,提出了一种结构自适应的自组织神经网络（ＳＡＳＯＮＮ）模型,ＳＡＳＯＮＮ基于把每个神经元看成是一个进化群体中的一个个体的观点,构造了神经元生长（ｇｒｏｗｉｎｇ）和删除（ｐｒｕｎｉｎｇ）的准则和方法,使得ＳＯＦＭ中的神经元欠利用,神经网络映射欠准确,以及映射的边缘效应等问题得到很大程度的改善。     

人工神经网络的发展及现状

人脑由复杂的神经网络构成,拥有极强的信息处理、学习和记忆能力,并且具有功耗低、神经元密度高等特点。利用现有的电路元件搭建出类似大脑功能的人工神经网络成为当今人工智能研究的热点。在人工神经网络的研究中,信息的编码、神经元模型的建立以及突触的选择是最关键的3个要素。分别介绍了4种不同种类的编码方式和4种不同类型的神经元模型。对忆阻器用作神经突触的特点进行了简要阐述,并对其数学模型和对应的工作方式作了详细概括。     

A Multiprocessor Architecture for the (M, L)-Algorithm Suitable for VLSI Implementation

The(M, L)-algorithm has been widely used in speech and image encoding. Recently, use of(M, L)-Iike algorithms has been suggested for decoding phase codes. With its ever-increasing use, there arises a need to explore architectures suitable for real-time applications. Toward this end, we present a multiprocessor architecture for the(M, L)algorithm that employs an SIMD (single instruction-multiple data) machine structure. The considerations involved in interconnection network design are discussed. The main functions of the network controller are switch state selection and synchronization. The number of switching elements required is significantly less than the elements required in the universal permutation network. These features make this architecture suitable for VLSI implementation. The tradeoff between number of processors and encoding time is also discussed.     

Event-driven simulation of integrate-and-fire models with spike-frequency adaptation

The evoked spike discharges of a neuron depend critically on the recent history of its electrical activity. A well-known example is the phenomenon of spike-frequency adaptation that is a commonly observed property of neurons. In this paper, using a leaky integrate-and-fire model that includes an adaptation current, we propose an event-driven strategy to simulate integrate-and-fire models with spike-frequency adaptation. Such approach is more precise than traditional clock-driven numerical integration approach because the timing of spikes is treated exactly. In experiments, using event-driven and clock-driven strategies we simulated the adaptation time course of single neuron and the random network with spike-timing dependent plasticity, the results indicate that (1) the temporal precision of spiking events impacts on neuronal dynamics of single as well as network in the different simulation strategies and (2) the simulation time scales linearly with the total number of spiking events in the event-driven simulation strategies.     

Biofunctional Silk/Neuron Interfaces

Silk fibroin (SF) is a biocompatible and slowly biodegradable material with excellent mechanical properties and huge potential for use as biofunctional interface in electronic devices that aim to stimulate and control neural network activity and peripheral nerve repair. It is shown that SF films act as material interfaces that support the adherence and neurite outgrowth of dorsal root ganglion (DRG) neurons and preserve neuronal functions. Silk films preserve the capability of neuronal cells to fire and DRG neurons on silk films retain the intracellular free Ca²⁺ concentration ([Ca²⁺]_i) response to capsaicin, a typical noxious stimulus for this neuronal culture model. It is also demonstrated that nerve growth factor (NGF)‐functionalized silk films promote neurite outgrowth and modulate functional properties of DRG neurons. The results show that silk preserves DRG neuronal physiology and is a promising biomaterial platform for the future development of devices with goals including functional recovery of injured neurons, neurite functional outgrowth in vitro, or functional electrostimulation in vivo.     

A surface-based 3-D dendritic spine detection approach from confocal microscopy images

Determining the relationship between the dendritic spine morphology and its functional properties is a fundamental challenge in neurobiology research. In particular, how to accurately and automatically analyse meaningful structural information from a large microscopy image data set is far away from being resolved. As pointed out in existing literature, one remaining challenge in spine detection and segmentation is how to automatically separate touching spines. In this paper, based on various global and local geometric features of the dendrite structure, we propose a novel approach to detect and segment neuronal spines, in particular, a breaking-down and stitching-up algorithm to accurately separate touching spines. Extensive performance comparisons show that our approach is more accurate and robust than two state-of-the-art spine detection and segmentation algorithms.     

A large-scale simulation of the piriform cortex by a cell automaton-based network model

An event-driven framework is used to construct a physiologically motivated large-scale model of the piriform cortex containing in the order of 10(5) neuron-like computing units. This approach is based on a hierarchically defined highly abstract neuron model consisting of finite-state machines. It provides computational efficiency while incorporating components which have identifiable counterparts in the neurophysiological domain. The network model incorporates four neuron types, and glutamatergic excitatory and GABA(A) and GABA(B) inhibitory synapses. The spatio-temporal patterns of cortical activity and the temporal and spectral characteristics of simulated electroencephalograms (EEGs) are studied. In line with previous experimental and compartmental work, 1) shock stimuli elicit EEG profiles with either isolated peaks or damped oscillations, the response type being determined by the intensity of the stimuli, and 2) temporally unpatterned input generates EEG oscillations supported by model-wide waves of excitation.     

Nonreciprocal Two-Ports Represented by Modified Wheeler Networks

The extension of the (reciprocal) modified Wheeler network to include the more general nonreciprocal two-port is given. This representation is derived via a known decomposition of the general nonreciprocal network into two portions, one reciprocal, the other nonreciprocal. The reciprocal portion is then taken as the modified Wheeler network. Recombination of the elements results in the desired representation which is constituted of a minimum number, i.e., of eight, passive elements. Each of these is a natural idealization of a physical microwave component. Since six of the elements belong to the class of "bilaterally matched" networks, some of the properties of this class are discussed. Two of the bilaterally matched elements embody the nonreciprocal properties of the network: a one-way attenuator and a one-way phase-shifter. Many of the characteristics of the (reciprocal) modified Wheeler network carry over directly to this nonreciprocal representation. The microwave measurement of the network parameters is also indicated.     

An analog neural computer with modular architecture for real-timedynamic computations

A multichip analog parallel neural network whose architecture, neuron characteristics, synaptic connections, and time constants are modifiable is described. The system has several important features, such as time constants for time-domain computations, interchangeable chips allowing a modifiable gross architecture, and expandability to any arbitrary size. Such an approach allows the exploration of different network architectures for a wide range of applications, in particular dynamic real-world computations. Four different modules (neuron, synapse, time constant, and switch units) have been designed and fabricated in a 2-μm CMOS technology. About 100 of these modules have been assembled in a fully functional prototype neural computer. An integrated software package for setting the network configuration and characteristics, and monitoring the neuron outputs has been developed as well. The performance of the individual modules as well as the overall system response for several applications was tested successfully. Results of a network for real-time decomposition of acoustical patterns are discussed     

激光陷阱引导神经细胞生长方向的研究

激光陷阱技术是近年来发展起来的一种非接触的操纵技术,它在生命科学领域取得了许多开创性成果,作者用特别设计的倒置式激光陷阱放在生长锥的前方来引导生长锥的生长方向,通过对生长锥施以持续的作用力,作者在实验上成功地引导了神经细胞生长锥的生长方向,并讨论了激光陷阱引导神经细胞生长的分子生物学基础。对这种新的引导神经细胞定向生长的方法的研究,可能对神经轴突的定向生长机制、控制神经再生产生非常积极的影响。     

Joining microelectronics and microionics: Nerve cells and brain tissue on semiconductor chips

The direct electrical interfacing of semiconductor chips with individual nerve cells and with brain tissue is considered. At first, the structure of the cell–chip contact is described and then the electrical coupling is characterized between ion channels, the electrical elements of nerve cells, and transistors and capacitors of silicon chips. On that basis, the signal transmission between microelectronics and microionics is implemented in both directions. Simple hybrid systems are assembled with neuron pairs and with small neuronal networks. Finally, the interfacing with capacitors and transistors is extended to brain tissue on silicon. The application of CMOS chips with capacitively coupled recording sites allows an imaging of neuronal activity with high spatiotemporal resolution. Goal of the work is an integration of neuronal network dynamics and digital electronics on a microscopic level for applications in brain research, medical prosthetics and information technology.     

电磁驱动下一类混合神经元模型的动力学响应与图像加密应用

在神经元活动的模型建立和分析过程中,应考虑一些生物物理效应。由于神经系统内部细胞内外离子浓度的波动,在集体电活动和神经元集群之间信号传播的过程中需要考虑电磁场的内部波动和跨膜磁通的影响。该文在一类混合神经元中引入磁通变量,通过对膜电位的调制诱发复杂的时变电磁场,运用Xppauto, Matcont和MATLAB等分析工具,探讨了新模型平衡点的存在性、初值敏感性和双参数分岔,发现外界刺激电流和电磁场变化时,可诱发新模型产生丰富的放电模式,如静息态、尖峰放电、周期(或混沌)簇放电,特别是由于磁通变量及忆阻器的引入产生的共存放电、隐藏放电等新现象。通过上述分析,基于电磁感应的神经元模型具有高非线性和较多的敏感参数,可使加密算法具有较大的密钥空间,基于此,该文设计了一种图像加密算法,对明文图像的像素先进行1次扩散再对其位置进行两次置乱。最后,通过一系列数值实验证明所设计的加密算法能有效地加密图像并且具有较高的安全性。该文考虑了神经细胞内外的电磁感应效应,有助于更全面了解神经元之间的信息编码和转迁规律,更多的分岔参数和高复杂性也使所设计的神经元模型在图像加密中具有很好的应用前景。     

Improving Pattern Discovery and Visualization of SAGE Data Through Poisson-Based Self-Adaptive Neural Networks

Serial analysis of gene expression (SAGE) allows a detailed, simultaneous analysis of thousands of genes without the need for prior, complete gene sequence information. However, due to its inherent complexity and the lack of complete structural and function knowledge, mining vast collections of SAGE data to extract useful knowledge poses great challenges to traditional analytical techniques. Moreover, SAGE data are characterized by a specific statistical model that has not been incorporated into traditional data analysis techniques. The analysis of SAGE data requires advanced, intelligent computational techniques, which consider the underlying biology and the statistical nature of SAGE data. By addressing the statistical properties demonstrated by SAGE data, this paper presents a new self-adaptive neural network, Poisson-based growing self-organizing map (PGSOM), which implements novel weight adaptation and neuron growing strategies. An empirical study of key dynamic mechanisms of PGSOM is presented. It was tested on three datasets, including synthetic and experimental SAGE data. The results indicate that, in comparison to traditional techniques, the PGSOM offers significant advantages in the context of pattern discovery and visualization in SAGE data. The pattern discovery and visualization platform discussed in this paper can be applied to other problem domains where the data are better approximated by a Poisson distribution.     

神经MOS晶体管

神经ＭＯＳ晶体管是１９９１年发明出来的一种具有高功能度的多输入栅控制的浮栅ＭＯＳ器件。本文介绍了它的基本结构和特点,论述了这种新哭喊 件及其电路的国外研究现状、研究趋以及笔者所完成的研究工作。     

INVESTIGATION ON THE COMPUTATIONAL PROPERTIES OF INTRANEURONAL DYNAMICS

This paper aims at exploring computational properties of dynamic processes in neu-ral systems,studying their mathematical formulation,and applying the results to artificial neuralnetwork modeling.The stimulus-response processes in neurons are first introduced briefly,thenproperties of neurons described by the Hodgkin-Huxley equations are analyzed.After studyinghow to simplify,the Hodgkin-Huxley equations while maintaining its properties,the concept of dy-namic neuron model is proposed.It is pointed out that the neuron model should include internalstates in order to obtain time-variant thresholds,such as refractory periods of neurons.Finallywe discuss problems related to neural network models based on pulse-stream communication andthe contribution of intraneuronal dynamics to collective properties of the neural network.