Characterizing the transmission of acupuncture signal: A combination of experimental and computational study

In order to reveal the underlying mechanisms about the conduction of acupuncture signal, manual acupuncture (MA) manipulations with different frequencies are taken at ‘Zusanli’ points of experimental rats. The induced electrical signal in spinal dorsal root ganglion is detected and recorded. It is found that rate coding features of acupuncture with different frequencies are different in PSTHs after spike sorting. Neuronal adaption and saturation phenomena are also observed. To elucidate the underlying mechanisms, we develop a delayed feedforward network model with plasticity for acupuncture signal transmission path. It is shown that this model could account for some important experimental phenomena and make further testable predictions. The underlying mechanisms of acupuncture are better understood by combining the experiment and simulation results.     

Active Disturbance Rejection Control of Quadrotor UAVs Based on Joint Observation and Feedforward Compensation北大核心CSCD

为解决模型参数不确定与外界干扰影响下,四旋翼无人机飞控作业中姿态与轨迹跟踪精度下降,反应迟缓的问题,利用拓展Kalman滤波应对非线性系统问题出色的适应能力和噪声抑制能力,对四旋翼状态信息进行初步估算来抑制高频信号干扰,从而降低了扩张状态观测器的估计负担.同时,与扩张状态观测器联合估计由系统不确定性参数与外界扰动联合组成的“总扰动”,使系统对于精确模型的依赖性降低,并利用扰动估计的微分值进行前馈补偿,以提高对突变扰动的跟踪精度,克服了突变干扰下的相位滞后现象.综合联合观测器、带前馈补偿的LESO及带误差补偿的PD控制律,形成了一种利用拓展Kalman滤波与前馈补偿后的扩张状态观测器联合观测扰动,能较大程度抑制高频噪声和突变扰动的改进型自抗扰控制器.仿真与实验结果表明,联合观测器能有效地减小观测误差幅值且能超前校正观测相位滞后,从而更好地得到更精确的状态信息,改进型自抗扰控制器能更好地满足四旋翼飞行器快速反应、高效稳定的控制要求,精准高效地完成复杂轨迹跟踪.     

Intelligent Crack Recognition Based on XFEM and GA-BP Neural Networks北大核心CSCD

基于扩展有限元法(XFEM)和经遗传算法(GA)优化的误差反向传播多层前馈(BP)神经网络(GA-BP)算法,建立了识别结构中裂纹的反演分析模型。模型通过XFEM正向分析获得的测点位移数据训练GA-BP神经网络,并在此基础上利用该网络进行裂纹反向识别。通过两个典型算例对模型的可行性和精度进行了验证,并探讨了网格密度、测点布置、输入数据噪声等对网络识别精度的影响。结果表明,该文的方法可反演线弹性断裂力学重点关注的直线裂纹的几何信息且具有较好的容噪性能,此外,GA-BP神经网络的预测精度较传统BP神经网络普遍更高。     

Nonlinear supratransmission in a discrete nonlinear electrical transmission line: Modulated gap peak solitons

We examine the propagation conditions of modulated gap signal voltage in a one-dimensional nonlinear discrete electrical transmission line where nonlinear capacitors are introduced both in their series and shunt branches. Especially, we consider that the propagating signal voltage frequency belong to the forbidden band zones. Using the exact discrete equations of the network and the extended nonlinear Schrödinger equation, the threshold value of the signal amplitude over which pulse soliton is emitted in the network is found. We show that when the signal amplitude exceeds a particular value, the critical value, peaked soliton can be emitted by the network due to the nonlinear dispersive elements of the series branches. Numerical simulations, confirming the exactness of the analytical analysis are performed on the exact equations of the network.     

Model selection in Neural Networks: Some difficulties

This paper considers two related issues regarding feedforward Neural Networks (NNs). The first involves the question of whether the network weights corresponding to the best fitting network are unique. Our empirical tests suggest an answer in the negative, whether using standard Backpropagation algorithm or our preferred direct (non-gradient-based) search procedure. We also offer a theoretical analysis which suggests that there will almost inevitably be functional relationships between network weights. The second issue concerns the use of standard statistical approaches to testing the significance of weights or groups of weights. Treating feedforward NNs as an interesting way to carry out nonlinear regression suggests that statistical tests should be employed. According to our results, however, statistical tests can in practice be indeterminate. It is rather difficult to choose either the number of hidden layers or the number of nodes on this basis.     

Optimal control of partially observed systems with arbitrary dependent noises: linear quadratic case

This paper deals with linear quadratic optimal control problem when signal and observation noises can be dependent. It is proved the separation principle for such case and it is shown how this separation principle enlarges the well-known duality between control and estimation problems. There are given existence results for three cases of the relations between signal and observation noises. One of them is a well-known independent noises case. Others concern the dependent noises.     

Comparing performance of feedforward neural nets and K-means for cluster-based market segmentation

We compare the performance of a specifically designed feedforward artificial neural network with one layer of hidden units to the K-means clustering technique in solving the problem of cluster-based market segmentation. The data set analyzed consists of usages of brands (product category: household cleaners) in different usage situations. The proposed feedforward neural network model results in a two segment solution that is confirmed by appropriate tests. On the other hand, the K-means algorithm fails in discovering any somewhat stronger cluster structure. Classification of respondents on the basis of external criteria is better for the neural network solution. We also demonstrate the managerial interpretability of the network results.     

Extended and Unscented Kalman filtering based feedforward neural networks for time series prediction

With the ability to deal with high non-linearity, artificial neural networks (ANNs) and support vector machines (SVMs) have been widely studied and successfully applied to time series prediction. However, good fitting results of ANNs and SVMs to nonlinear models do not guarantee an equally good prediction performance. One main reason is that their dynamics and properties are changing with time, and another key problem is the inherent noise of the fitting data. Nonlinear filtering methods have some advantages such as handling additive noises and following the movement of a system when the underlying model is evolving through time. The present paper investigates time series prediction algorithms by using a combination of nonlinear filtering approaches and the feedforward neural network (FNN). The nonlinear filtering model is established by using the FNN’s weights to present state equation and the FNN’s output to present the observation equation, and the input vector to the FNN is composed of the predicted signal with given length, then the extended Kalman filtering (EKF) and Unscented Kalman filtering (UKF) are used to online train the FNN. Time series prediction results are presented by the predicted observation value of nonlinear filtering approaches. To evaluate the proposed methods, the developed techniques are applied to the predictions of one simulated Mackey-Glass chaotic time series and one real monthly mean water levels time series. Generally, the prediction accuracy of the UKF-based FNN is better than the EKF-based FNN when the model is highly nonlinear. However, comparing from prediction accuracy and computational effort based on the prediction model proposed in our study, we draw the conclusion that the EKF-based FNN is superior to the UKF-based FNN for the theoretical Mackey-Glass time series prediction and the real monthly mean water levels time series prediction.     

非平衡拓扑和随机干扰情形下多自主体系统的趋同条件

研究了具有一般有向通信拓扑和高斯通信噪声的多自主体系统的趋同条件.这里所研究的有向拓扑不仅包含有向平衡图,而且包含非平衡图,后者是本文的重点.我们利用马氏链的结果得到了一个网络节点的互通类;通过对噪声影响的细化,给出了不同噪声情形下系统趋同条件:(1)对互通类的自主体获取信息受到噪声干扰情形,给出了系统均方趋同的充要条件,并证明该条件也保证以概率1 趋同;(2)对互通类的自主体获取信息未受到噪声干扰但其余自主体获取信息受到干扰情形,给出了系统均方趋同的充分条件,并证明该条件在一定意义下也是必要的;(3)对整个系统无噪声情形,给出了系统趋同的充要条件.     

Modeling environmental effects on directionality in wireless networks

One method for improving wireless network throughput involves using directional antennas to increase signal gain and/or decrease interference. The physical layer models used in current networking simulators only minimally address the interaction of directional antennas and radio propagation. This paper compares the models found in popular simulation tools with measurements taken across a variety of links in multiple environments. We find that the effects of antenna direction are significantly different from those predicted by the models used in the common wireless network simulators. We propose a parametric model that better captures the effects of different propagation environments on directional antenna systems; we also show that the derived models are sensitive to both the direction of signal gain and the environment in which the antenna is used.     

Development of spiral wave in a regular network of excitatory neurons due to stochastic poisoning of ion channels

Some experimental evidences show that spiral wave could be observed in the cortex of brain, and the propagation of this spiral wave plays an important role in signal communication as a pacemaker. The profile of spiral wave generated in a numerical way is often perfect while the observed profile in experiments is not perfect and smooth. In this paper, formation and development of spiral wave in a regular network of Morris–Lecar neurons, which neurons are placed on nodes uniformly in a two-dimensional array and each node is coupled with nearest-neighbor type, are investigated by considering the effect of stochastic ion channels poisoning and channel noise. The formation and selection of spiral wave could be detected as follows. (1) External forcing currents with diversity are imposed on neurons in the network of excitatory neurons with nearest-neighbor connection, a target-like wave emerges and its potential mechanism is discussed; (2) artificial defects and local poisoned area are selected in the network to induce new wave to interact with the target wave; (3) spiral wave can be induced to occupy the network when the target wave is blocked by the artificial defects or poisoned area with regular border lines; (4) the stochastic poisoning effect is introduced by randomly modifying the border lines (areas) of specific regions in the network. It is found that spiral wave can be also developed to occupy the network under appropriate poisoning ratio. The process of growth for the poisoned area of ion channels poisoning is measured, the effect of channels noise is also investigated. It is confirmed that perfect spiral wave emerges in the network under gradient poisoning even if the channel noise is considered.     

Spatial distribution complexities of traffic congestion and bottlenecks in different network topologies

Recently, urban traffic congestion has become a popular social problem. The generation and the propagation of congestion has close relation with the network topology, the traffic flow, etc. In this study, based on the traffic flow propagation method, we investigate the time and space distribution characteristics of the traffic congestion and bottlenecks in different network topologies (e.g., small world, random and regular network). The simulation results show that the random network is an optimal traffic structure, in which the traffic congestion is smaller than others. Moreover, the regular network is the worst topology which is prone to be congested. Additionally, we also prove the effects of network with community structure on the traffic system and congestion bottlenecks including its generation, propagation and time–space complexities. Results indicate that the strong community structure can improve the network performance and is effective to resist the propagation of the traffic congestion.     

A neural fuzzy control system with structure and parameter learning

A general connectionist model, called neural fuzzy control network (NFCN), is proposed for the realization of a fuzzy logic control system. The proposed NFCN is a feedforward multilayered network which integrates the basic elements and functions of a traditional fuzzy logic controller into a connectionist structure which has distributed learning abilities. The NFCN can be constructed from supervised training examples by machine learning techniques, and the connectionist structure can be trained to develop fuzzy logic rules and find membership functions. Associated with the NFCN is a two-phase hybrid learning algorithm which utilizes unsupervised learning schemes for structure learning and the backpropagation learning scheme for parameter learning. By combining both unsupervised and supervised learning schemes, the learning speed converges much faster than the original backpropagation algorithm. The two-phase hybrid learning algorithm requires exact supervised training data for learning. In some real-time applications, exact training data may be expensive or even impossible to obtain. To solve this problem, a reinforcement neural fuzzy control network (RNFCN) is further proposed. The RNFCN is constructed by integrating two NFCNs, one functioning as a fuzzy predictor and the other as a fuzzy controller. By combining a proposed on-line supervised structure-parameter learning technique, the temporal difference prediction method, and the stochastic exploratory algorithm, a reinforcement learning algorithm is proposed, which can construct a RNFCN automatically and dynamically through a reward-penalty signal (i.e., “good” or “bad” signal). Two examples are presented to illustrate the performance and applicability of the proposed models and learning algorithms.     

小波分析在信号去噪声中的应用

小波分析是近年来发展起来的一种数学方法,在信号与图象处理中有重要的应用.中值滤波是信号处理中常用的一种非线性滤波器,它能够有效地消除瞬时脉冲干扰,并且能够很好地保持信号的边缘信息,在信号和图象处理中得到广泛应用.对中值滤波器与小波变换的结合进行了比较系统的研究.通过实例说明中值滤波器与小波变换相结合具有比单一滤波器更好的效果.     

An Electrocardiogram Signal Classification Algorithm Based on Improved Deep Residual Shrinkage Networks北大核心CSCD

心电信号分类是医疗保健领域的重要研究内容.针对大多数方法不能很好地降低样本数量少的类别漏诊率,以及降低预处理操作的复杂性问题,提出了一种基于改进深度残差收缩网络(IDRSN)的心电信号分类算法(即DRSL算法).首先,使用合成少数类过采样技术(SMOTE)扩充数量少的类别样本,从而解决了类不平衡问题;其次,利用改进深度残差收缩网络提取空间特征,其残差模块可以避免网络层加深造成的过拟合,压缩激励和软阈值化子网络可以提取重要局部特征并自动去除噪声;然后,通过长短期记忆网络(LSTM)提取时间特征;最后,利用全连接网络输出分类结果.在MIT-BIH心律失常数据集上的实验结果表明,该算法的分类性能优于IDRSN、DRSN、GAN+2DCNN、CNN+LSTM_ATTENTION、SE-CNN-LSTM分类算法.     

Stochastic resonance in a linear static system driven by correlated multiplicative and additive noises

In this work, we investigate the signal transmission in a linear static system driven by correlated multiplicative and additive noises. When the input signal is periodic, we depict the stochastic resonance (SR) phenomenon by employing the signal-to-noise ratio (SNR) theory; while the input signal is aperiodic, we describe the SR phenomenon by using the input–output cross correlation theory. And the exact analytic expressions of the output SNR and the normalized time averaged cross covariance between input and output are obtained. The results show: under the condition of negative correlated noises, SR arises; while with positive correlated or uncorrelated noises, there is no SR. This result may extend the SR theory to a common linear static system.     

Multifractal detrended fluctuation analysis of analog random multiplicative processes

We investigate non-Gaussian statistical properties of stationary stochastic signals generated by an analog circuit that simulates a random multiplicative process with weak additive noise. The random noises are originated by thermal shot noise and avalanche processes, while the multiplicative process is generated by a fully analog circuit. The resulting signal describes stochastic time series of current interest in several areas such as turbulence, finance, biology and environment, which exhibit power-law distributions. Specifically, we study the correlation properties of the signal by employing a detrended fluctuation analysis and explore its multifractal nature. The singularity spectrum is obtained and analyzed as a function of the control circuit parameter that tunes the asymptotic power-law form of the probability distribution function.     

三层前向神经网络的一致逼近性

近年来，前向神经网络泛逼近的一致性分析一直为众多学者所重视。本文系统分析三层前向网络对于拟差值保序函数族的一致逼近性，其中，转换函数σ是广义Sigmoidal函数。并将此一致性结果用于建立一类新的模糊神经网络(FNN)，即折线FNN．研究这类网络对于两个给定的模糊函数的逼近性，相关结论在分析折线FNN的泛逼近性时起关键作用。