一种基于深度学习的心电图分类方法

心血管疾病是导致人类死亡的主要原因之一,在心血管疾病的诊断上,传统的医学方法需要有大量知识储备的专业人员,在高素质从医人员较少的欠发达地区易出现误诊的现象。为了节约医疗资源,降低心血管疾病误诊率,基于历史医疗数据,提出一种基于经验模态分解(empirical mode decomposition,EMD)与人工神经网络(artificial neural network,ANN)的新型面向心血管疾病数据分类模型,在分类模型中引入注意力机制(SE-Net模块),提高了检测效率。将所提出模型试用于多个公开数据集上,在MIT-BIH心律不齐数据库、欧洲ST-T心电数据库上精准率均表现良好。实验结果表明:在数据去噪任务中,文章提出模型的去噪性能明显优于基准模型;在疾病分类任务中,所提出模型的精准率达到95.78%,召回率达到86.16%,优于目前主流分类模型。同时,还设计了消融实验,用于验证两个模块的必要性。消融实验结果显示,文中两个模块对分类的准确性和鲁棒性均有提升,具有充分的必要性。

软错误缓解动态部分可重构的抗单粒子方案

随着基于静态随机存储器(static random-access memory,SRAM)型现场可编程门阵列(field programmable gate array,FPGA)广泛应用于航空航天领域,太空辐照环境下FPGA产生单粒子翻转(single event upset, SEU)问题的概率日益提高,从而导致FPGA出现单粒子闩锁现象引起功能紊乱。针对该问题,基于SRAM型FPGA的架构诱发SEU机理分析,对传统三模冗余(triple module redundancy, TMR)的方案进行改进,设计一种逻辑上采用TMR进行备份、系统上采用软错误算法缓解执行,同时采用局部纠错和动态可重构的方法进行抑制的方案。皮秒激光注入试验结果显示,采用所提供方案的FPGA较传统方案试验电流平稳,验证了该方案可以有效对SEU进行抑制。

复杂网络牵制控制概述

本文回顾了十多年来复杂网络牵制控制的研究进展,对复杂网络现有的牵制控制算法进行了归纳分类,然后基于网络拓扑及节点动力学特性讨论了网络牵制控制的一些必要或充分条件.针对牵制控制下的复杂网络,重点论述了牵制节点的选取策略,综合分析了牵制节点个数、耦合强度、牵制控制增益及节点动力学等因素对复杂网络牵制控制的影响,讨论了具有合作竞争关系符号网络牵制控制的进展,并简要介绍了复杂网络牵制控制的一些应用前景.最后,就复杂网络牵制控制将来的研究方向提了一些建议及展望.

基于指数Laplace损失函数的回归估计鲁棒超限学习机

实际问题的数据集通常受到各种噪声的影响,超限学习机(extreme learning machine,ELM)对这类数据集进行学习时,表现出预测精度低、预测结果波动大.为了克服该缺陷,采用了能够削弱噪声影响的指数Laplace损失函数.该损失函数是建立在Gauss核函数基础上,具有可微、非凸、有界且能够趋近于Laplace函数的特点.将其引入到超限学习机中,提出了鲁棒超限学习机回归估计(exponential Laplace loss function based robust ELM for regression,ELRELM)模型.利用迭代重赋权算法求解模型的优化问题.在每次迭代中,噪声样本点被赋予较小的权值,能够有效地提高预测精度.真实数据集实验验证了所提出的模型相比较于对比算法具有更优的学习性能和鲁棒性.

政务服务中群众留言答复意见评价模型

为了提高政府事务管理水平和处理效率,打造“智慧政务”平台,针对政务服务中群众留言的答复内容进行客观评价,提出了一种结合多算法的群众留言回复评价模型.首先,根据答复文本内容定义答复长度、相似性、完整性、可解释性和及时性5个评价特征指标,分别从内容、格式、合理程度和时间4个角度对文本进行评价;其次,采用回归分析方法对答复意见类型进行分析;接着,分别采用K-means聚类、DBSCAN聚类算法、Meanshift聚类算法、HC聚类算法对答复意见等级进行分析,根据4种聚类算法效果对比结果,选用K-means聚类算法,再将聚类算法和回归分析的结果结合,对答复意见进行评价,并将评价等级分为6类.本文模型将机器学习中的数据挖掘和数据分析算法与“智慧政务”相结合,为政府事务管理的研究提供了一个量化分析的工具.

Convergence in Continuous Hopfield Neural Network with Delays

A sufficient condition are derived for the global asymptotic stability of the equilibrium of continuous Hopfield neural networks with delays of the

一类具有迁移和Allee效应的食饵-捕食者系统稳定性

研究一类食饵具有Allee效应且捕食者具有人工控制迁移的食饵-捕食者系统,该系统具有平方根项的功能性反应函数。首先通过定性分析,证明解的有界性,分析平衡点的存在性,得到系统平衡点的局部稳定性的充分条件。接着讨论平衡点的Hopf分岔存在性,并通过计算第一李雅普诺夫系数,研究平衡点Hopf分岔的稳定性和方向。最后通过数值模拟验证所得结论的正确性,结果表明Allee效应和人工控制迁移率对食饵种群和捕食者种群的生存与灭绝具有重要意义。

一种领队车输入未知的多车辆纵向协作控制方法

本文考虑了领队车控制输入未知的情况,提出一种多车辆纵向协作控制方法.首先,通过采用精确反馈线性化技术,得到了线性的车辆的动力学模型.然后,采用双向领队跟随通信策略,基于邻居车辆的状态信息,为每辆跟随车设计分布式控制律.考虑在领队车输入有界的情况下,提出了一种有效的多车辆协作控制算法,能够保证多车辆系统以最大的收敛速率达到内部稳定.最后,仿真结果展示了所提出控制器设计算法的有效性和优越性.

Distributed online bandit tracking for Nash equilibrium under partial-decision information setting

This paper is concerned with a Nash equilibrium(NE)tracking issue in online games with bandit feedback,where cost functions vary with time and agents only have access to the values of these functions at two points during each round.A partial-decision information setting is considered,in which agents have only access to the decisions of their neighbors.The primary objective of this paper is to develop a distributed online NE tracking algorithm that ensures sublinear growth of regret with respect to the total round T,under both the bandit feedback and partial-decision information setting.By utilizing a two-point estimator together with the leader-following consensus method,a new distributed online NE tracking algorithm is established with the estimated gradient and local estimated decisions based on the projection gradient-descent method.Moreover,sufficient conditions are derived to guarantee an improved upper bound of dynamic regret compared to existing bandit algorithms.Finally,a simulation example is presented to demonstrate the effectiveness of the proposed algorithm.

Bipartite quasi-synchronization of multiple neural networks with generalized cooperative-competitive topology

This study examines the bipartite quasi-synchronization(B-Q synchronization)issue of coupled networks with general cooperative-competitive topology and the event-triggered communications between nodes to curb the communication cost.In the existing literature concerning bipartite synchronization,the network topology is required to be structurally balanced,which necessitates that competitive interactions exist only between two distinct subgroups.In this study,we aim to lengthen the network’s topology to a more general signed network in which antagonistic interactions can exist in the same or different subgroups.According to signed graph theory and the markovian stochastic event-triggering mechanism,the dynamical model of multiple neural networks(MNNs)with structurally unbalanced and markovian event-triggered communication is established.By utilizing the stochastic Lyapunov stability analysis,some adequate criteria for B-Q synchronization of MNNs with the structurally unbalanced graph are obtained;also,a bound for the B-Q synchronization error is provided.As a special case,the bipartite synchronization criteria for MNNs with the structurally balanced graph are also obtained.Finally,two simulations are performed to verify the theoretical result.

非真实感绘制技术研究现状与展望

非真实感绘制技术(non-photorealistic rendering,NPR)主要用于模拟艺术风格、表现艺术特质和传达用户情感等,是计算机图形学的重要组成部分,其研究对象逐渐丰富,研究方法不断创新。本文从基于图像建模的绘制方法、基于深度学习的绘制方法、中国特有艺术作品的数字化模拟、非真实感情感特征识别以及非真实感视频场景绘制等5个方面概述目前研究进展,然后从扩展非真实感研究对象、增强视频绘制帧间连贯性、提取艺术风格情感特征以及评价非真实感绘制结果等4个角度讨论需要进一步研究的问题。针对需要深入研究的问题,指出提高算法的通用性和绘制效率,以及提高深度学习网络的泛化性,有助于扩展研究对象,模拟艺术风格的多样性,同时减小视频场景的帧间跳变;对艺术风格作品具有的情感特征、内在机理特征进行模拟,有助于提高绘制结果与艺术风格图像的相似度;结合主观和客观评价模型,可以更准确地对绘制结果进行评价,同时有利于优化网络模型参数,提高绘制效率。非真实感绘制在计算机视觉、文化遗产保护等领域具有重要的应用前景,但其研究对象、绘制算法、绘制效率仍然存在很多亟待解决的问题,随着硬件设备的不断改进,综合运用学科交叉知识、扩展应用领域将进一步推动非真实感绘制技术的发展。

Exponential H_∞ filtering analysis for discrete-time switched neural networks with random delays using sojourn probabilities

This paper is concerned with the exponential H_∞ filtering problem for a class of discrete-time switched neural networks with random time-varying delays based on the sojourn-probability-dependent method. Using the average dwell time approach together with the piecewise Lyapunov function technique, sufficient conditions are proposed to guarantee the exponential stability for the switched neural networks with random time-varying delays which are characterized by introducing a Bernoulli stochastic variable.Based on the derived H_∞ performance analysis results, the H_∞ filter design is formulated in terms of Linear Matrix Inequalities(LMIs). Finally, two numerical examples are presented to demonstrate the effectiveness of the proposed design procedure.

Finite-time stabilization of switched dynamical networks with quantized couplings via quantized controller

This paper concerns the stabilization of switched dynamical networks with logarithmic quantization couplings in a settling time.The switching sequence is constrained by hybrid dwell time. Controller is designed by using limited information. Due to the quantization and switching, traditional finite-time analysis methods cannot be utilized directly. By designing multiple Lyapunov functions and constructing comparison systems, a general criterion formulated by matrix inequalities is first given. Then specific conditions in terms of linear matrix inequalities are established by partitioning the dwell time and using convex combination technique. An optimal algorithm is proposed for the estimation of settling time. Numerical simulations are given to verify the effectiveness of the theoretical results.

Synchronization criteria for multiple memristor-based neural networks with time delay and inertial term

This present work uses different methods to synchronize the inertial memristor systems with linear coupling. Firstly, the mathematical model of inertial memristor-based neural networks(IMNNs) with time delay is proposed, where the coupling matrix satisfies the diffusion condition, which can be symmetric or asymmetric. Secondly, by using differential inclusion method and Halanay inequality, some algebraic self-synchronization criteria are obtained. Then, via constructing effective Lyapunov functional, designing discontinuous control algorithms, some new sufficient conditions are gained to achieve synchronization of networks. Finally, two illustrative simulations are provided to show the validity of the obtained results, which cannot be contained by each other.

Selecting pinning nodes to control complex networked systems

One of the fundamental problems in pinning control of complex networks is selecting appropriate pinning nodes, such that the whole system is controlled. This is particularly useful for complex networks with huge numbers of nodes. Recent research has yielded several pinning node selection strategies, which may be efficient. However, selecting a set of pinning nodes and identifying the nodes that should be selected first remain challenging problems. In this paper, we present a network control strategy based on left Perron vector. For directed networks where nodes have the same in-and out-degrees, there has so far been no effective pinning node selection strategy, but our method can find suitable nodes. Likewise, our method also performs well for undirected networks where the nodes have the same degree. In addition, we can derive the minimum set of pinning nodes and the order in which they should be selected for given coupling strengths. Our proofs of these results depend on the properties of non-negative matrices and M-matrices. Several examples show that this strategy can effectively select appropriate pinning nodes, and that it can achieve better results for both directed and undirected networks.

CPSO-XGBoost segmented regression model for asphalt pavement deflection basin area prediction

The use of non-destructive testing(NDT) equipment, such as the falling weight deflectometer(FWD), provides important estimates of road health and helps to optimize road management regimes. However, periodic road testing and post-processing of the collected data are cumbersome and require much expertise, a considerable amount of time, money, and other resources. This study attempts to develop a reliable prediction method for estimating the deflection basin area of different asphalt pavements using road temperature, load time, and load pressure as main characteristics. The data are obtained from 19 kinds of asphalt pavements on a 2.038-km-long full-scale fleld accelerated pavement testing track named RIOHTrack(Research Institute of Highway Track) in Tongzhou, Beijing. In addition, a chaotic particle swarm algorithm(CPSO) and a segmented regression strategy are proposed in this paper to optimize the XGBoost model. The experiment results of the proposed method are compared with those of classical machine learning algorithms and achieve an average of mean square error and mean absolute error respectively by 5.80 and 1.59.The experiments demonstrate the superiority of the XGBoost algorithm over classical machine learning methods in dealing with nonlinear problems in road engineering. Signiflcantly, the method can reduce the frequency of deflection tests without affecting its estimation accuracy, which is a promising alternative way to facilitate the rapid assessment of pavement conditions.

Comparative study on bifurcation control methods in a fractional-order delayed predator-prey system

This paper focuses on the comparative analysis of bifurcation control approaches for a fractional-order delayed predator-prey system. The state feedback schemes with and without time delay are dexterously designed during the bifurcation control for the proposed system, and the comparative study is elaborately performed on bifurcation control theoretically. The salient feature of the current paper is that the analysis of time of bifurcation control for the proposed system is investigated based on various state feedback control strategies. Analysis reveals that the feedback control approach without time delay overly outmatches the one with time delay for bifurcation control in the considered systems so long as the uniform feedback gain is selected. It consumes less time to control bifurcation while utilizing the first controller in comparison with the second one. Numerical examples are ultimately employed to confirm the correctness of the theoretical results.

Research on the basic theoretical issues of smart metro construction

The basic conception and theoretical system of smart metro are not clear and there is not a unified understanding at the academic level.Based on this fact,there is no guidance of the smart metro construction.Therefore,this paper conducts a complete and systemic research on several basic theoretical issues of smart metro,such as the development history,concept,connotation,characteristics,evolution stages,system architecture and main construction content.Firstly,the internal and external development demands are analyzed in this paper.In addition,the subject-cycle-function(SCF)model of smart metro concept is obtained.Moreover,the essential connotation,three typical characteristics and the evolution process of smart metro are introduced.Furthermore,the evolution relationships of main function,achievement goal,instrument subject,decision subject,control subject and factor relationship for every evolution stage are clarified.Finally,according to the theory of system engineering,the three-hierarchy system architecture and five main construction contents are proposed.The both can be used to construct the technical system of smart metro at the academic level,and provide theoretical support and technical guidance for promoting the development of smart metro construction in China.

Event-triggered bipartite consensus over cooperation-competition networks under DoS attacks

This paper addresses the bipartite consensus over cooperation-competition networks affected by denial-of-service(DoS) attacks.Consider that a network consists of multiple interactive agents, and the relationship between neighboring agents is cooperative or competitive. Meanwhile, information transmission among the agents is unavailable during the intervals of attacks. In order to save communication resources and exclude the Zeno behavior, an event-triggered scheme depending on the sampled-data information from neighboring agents is proposed, and efficient defense strategies in response to the attacks are put forward.Suppose that the frequency and duration of DoS attacks meet certain requirements, then according to the signed graph theory,the La Salle’s invariance principle, and the convergence of monotone sequences, the results of bipartite consensus via the eventtriggered protocol are provided, which are mainly related to the communication topology of the network, the sampling period,and the threshold parameters in the event-triggered scheme. It is shown that the bipartite consensus is realized even though the DoS attacks take place frequently. Furthermore, this paper discusses the bipartite consensus in the presence of DoS attacks with a random unsuccessful rate. Finally, numerical simulations illustrate the theoretical results.

Traffic network equilibrium problems with demands uncertainty and capacity constraints of arcs by scalarization approaches

This paper focuses on the vector traffic network equilibrium problem with demands uncertainty and capacity constraints of arcs, in which, the demands are not exactly known and assumed as a discrete set that contains finite scenarios. For different scenario, the demand may be changed, which seems much more reasonable in practical programming. By using the linear scalarization method,we introduce several definitions of parametric equilibrium flows and reveal their mutual relations. Meanwhile, the relationships between the scalar variational inequality as well as the(weak) vector equilibrium flows are explored, meanwhile, some necessary and sufficient conditions that ensure the(weak) vector equilibrium flows are also considered. Additionally, by means of nonlinear scalarization functionals, two kinds of equilibrium principles are derived. All of the derived conclusions contain the demands uncertainty and capacity constraints of arcs, thus the results proposed in this paper improved some existing works. Finally, some numerical examples are employed to show the merits of the improved conclusions.