CONVERGENCE OF ONLINE GRADIENT METHOD WITH A PENALTY TERM FOR FEEDFORWARD NEURAL NETWORKS WITH STOCHASTIC INPUTS

Online gradient algorithm has been widely used as a learning algorithm for feedforward neural network training. In this paper, we prove a weak convergence theorem of an online gradient algorithm with a penalty term, assuming that the training examples are input in a stochastic way. The monotonicity of the error function in the iteration and the boundedness of the weight are both guaranteed. We also present a numerical experiment to support our results.     

Characterizing neural activities evoked by manual acupuncture through spiking irregularity measures

The neural system characterizes information in external stimulations by different spiking patterns. In order to examine how neural spiking patterns are related to acupuncture manipulations, experiments are designed in such a way that different types of manual acupuncture （MA） manipulations are taken at the ＇Zusanli＇ point of experimental rats, and the induced electrical signals in the spinal dorsal root ganglion are detected and recorded. The interspike interval （ISI） statistical histogram is fitted by the gamma distribution, which has two parameters： one is the time-dependent firing rate and the other is a shape parameter characterizing the spiking irregularities. The shape parameter is the measure of spiking irregularities and can be used to identify the type of MA manipulations. The coefficient of variation is mostly used to measure the spike time irregularity, but it overestimates the irregularity in the case of pronounced firing rate changes. However, experiments show that each acupuncture manipulation will lead to changes in the firing rate. So we combine four relatively rate- independent measures to study the irregularity of spike trains evoked by different types of MA manipulations. Results suggest that the MA manipulations possess unique spiking statistics and characteristics and can be distinguished according to the spiking irregularity measures. These studies have offered new insights into the coding processes and information transfer of acupuncture.     

N型局部有源忆阻器的神经形态行为

局部有源忆阻器(locally-active memristor,LAM)凭借其高集成度、低功耗和局部有源特性等优点,在神经形态计算领域显示出巨大的潜力.本文提出了一种简单的N型LAM数学模型,通过揭示其非线性动力特性,设计了N型LAM神经元电路.采用Hopf分岔、数值分析等方法定量研究了该电路的动力学行为,成功模拟了多种神经形态行为,包括全或无行为、尖峰、簇发、周期振荡等.并利用该神经元电路结构模拟了生物触觉神经元的频率特性.仿真结果表明:当输入信号幅值低于阈值时,神经元电路输出信号的振荡频率与输入信号强度呈正相关(即兴奋状态),并在阈值处达到最大值.随后,继续增大激励强度,振荡频率则逐渐降低(即保护性抑制状态).最后,设计了N型LAM硬件仿真器,并完成了人工神经元电路的硬件实现,实验结果与仿真结果、理论分析相一致,验证了该N型LAM具备的神经形态行为.     

具有感存算一体化的新型神经形态视觉传感器

传统的数字图像处理系统包括图像传感器与图像处理单元,二者在物理空间上分离,图像信息在其间的传输造成了延时与能耗。此外,数字图像传感器基于“帧”的工作原理,可能丢失一些重要信息,或者造成数据冗余。人类视觉系统提供了一种高效并行的信息处理方式。神经形态视觉传感器能够模拟人类视网膜的功能,同时具备感知光信号、存储信号和进行信息预处理的功能。这类感存算一体化的神经形态视觉传感器简化了人工视觉系统的电路复杂性,提升了信息处理效率,节省了系统功耗。文章总结了传统的数字图像传感器存在的问题,介绍了几种重要的人工神经网络,讨论了新型神经形态视觉传感器的研究进展和存在的问题。     

基于样本熵的听觉神经锁相机理的实验分析

锁相是指系统的响应与周期性刺激的特定相位同步的物理现象. 听觉神经的锁相对揭示人的听觉认知基本的神经机理及改善听觉感知有重要意义. 然而, 现有研究主要集中于心理物理方法和幅度谱分析, 不能有效区分包络响应和时域细节结构响应, 不能直观反映神经锁相. 本文主要利用拔靴法和离散傅里叶变换, 提出了基于样本熵的时域细节结构频率跟随响应(temporal-fine-structure-related frequency following response, FFR_T)的神经锁相值(phase locking value, PLV)计算方法, 用于分析神经物理实验数据. 两个脑电实验结果表明: FFR_T的PLV样本熵显著大于包络相关频率跟随响应(envelope-related frequency following response, FFR_E)的PLV, 且二者正交独立, 新方法能有效地分别反映听觉系统对包络和时间细节结构的锁相机理; 基频处的响应主要来源于FFR_E的锁相; 基频处, 不可分辨谐波成分包络的锁相能力优于对可分辨谐波; 基频缺失时, 畸变产物是不同的听觉神经通路的FFR_E的混合; 谐波处, FFR_E 集中于低频, FFR_T则集中于中、高频; 听觉神经元锁相能力与声源的频率可分辨性相关. FFR_T的PLV方法克服了现有FFR分析的局限性, 可用于深入研究听觉神经机理.     

老年痴呆治愈有望

<正>许多神经退行性疾病,包括老年痴呆,都是蛋白质产生错误折叠或出现其他差错(称为朊病毒)所致。作为对抗朊病毒的防御机制的一部分,大脑会发现病毒蛋白并停止蛋白质的生产,以防止病毒扩散。这一机制对其他错误蛋白也会启动,但停止生产蛋白质会导致脑细胞的死亡。     

连续臂丛神经阻滞镇痛对断指再植血流动力学影响的研究

目的研究连续臂丛神经阻滞镇痛对断指再植术后再植指血流动力学的影响,探讨其缓解血管痉挛、增加血流量的可行性。方法将60例急诊行断指再植术患者按随机数字表法分为术后连续臂丛神经阻滞镇痛组（CA组）和对照组（NA组）,每组30例。观察两组患者再植指术后血管痉挛、血管栓塞和指体坏死发生例数并计算再植指成活率,记录两组术后即刻（T1）、术后24h（T2）再植指固有动脉血流动力学参数及动脉内径（AD）、远指端血氧饱和度（SpO2）和温度（Ts）变化。结果断指再植术后成活率CA组高于NA组,血管痉挛、血管栓塞和指体坏死发生率CA组均低于NA组,差异均有统计学意义（均P＜0.05）。在T2时CA组各项血流动力学参数、AD、SpO2、Ts均高于NA组,阻力指数（RI）低于NA组,差异均有统计学意义（均P＜0.05）。结论术后连续臂丛神经阻滞镇痛可以缓解再植指固有动脉痉挛,增加指体血流量,提高了断指再植的成功率。     

Comparison of Mechanisms of N-Nitrosation and N-Nitration of Ammonia and Dimethylamine by Reactive Nitrogen Oxygen Species: A Theoretical Study

Reactive nitrogen oxygen species(RNOS) implicate damage in biological systems,especially leading to inflammation,neurodegenerative and cardiovascular diseases,and cancer by altering the functions of biomolecules through the N-nitrosation and N-nitration reactions.The mechanisms of N-nitrosation and N-nitration reactions of ammonia and dimethylamine by RNOS,i.e.,N2O3,N2O4,N2O5 and ONOOH,were investigated at the CBS-QB3 level of theory.The computational results indicate that the N-nitrosation reaction prefers a concerted mechanism,in which a H-abstraction and ON-addition occur simultaneously,whereas a stepwise mechanism(also called a free radical mechanism) is more favorable for most nitrating agents in the N-nitration reaction,where NO2 first abstracts a hydrogen atom from the nitrogen of amines and then the induced intermediate reacts with NO2 once more to form the nitration products.However,the concerted pathway is still a feasible process for some nitrating agents such as N2O5.In addition,the relationship between the structures of different RNOS and their nitrosating or nitrating abilities was also investigated.