网络结构特征与链路预测算法关系研究

以美国航空网络、科学家合作网络和线虫新陈代谢网络等5种实际网络为例进行了综合实验,用结果数据定量化描述了同配系数、集聚系数和网络效率等网络结构特征参数,与基于局部信息和全局信息的两类链路预测方法结果之间的关系。通过对结果的分析,得到了网络同配系数为正且聚集系数大于阈值(约0.1)时适用基于局部信息的预测方法,否则适用基于全局信息的预测方法;以及集聚系数、网络效率与局部信息预测方法的结果成正比,与全局信息预测方法成反比等结论。这些结论为通过网络特征参数进行链路预测方法的选择提供了定量化的参考依据。     

多层同步网络在舆情仿真研究中的应用

为统筹考虑线上社交网络与线下社交网络在舆情传播中的作用,首次提出了一种包含"媒体层-线上层-线下层"的多层同步网络模型,并在此基础上搭建了舆情仿真系统框架.基于对人类社交网络拓扑结构的已有研究成果,给出了多层同步网络的缺省设定.通过案例仿真,分析了线上网络与线下网络之间的相互影响,对比了舆情事件目击者数量在多媒体传播、自媒体传播这两种舆论传播模式中的作用差异.最后,讨论了多层同步网络模型的兼容性与可扩展性,并指出了该模型改进的方向.     

媒体负面报道对公司股价影响的研究:来自白酒塑化剂事件的证据

金融学理论认为,资本价格会因意外重要事件的冲击而发生变化.过去许多文献讨论过股票价格与媒体报道这一外部冲击之间是否存在显著关系,但是,这些研究所得结论并不尽一致.基于此,本文以白酒塑化剂事件为例,分析塑化剂曝光新闻数量对白酒公司股价的影响.研究表明,塑化剂曝光在较长时间内影响了白酒板块的股价,但是曝光事件对各个公司股价影响的程度不一.此外,研究还发现,中央反腐抑制了白酒板块的股价.     

INTELLIGENT SECURITY SYSTEMS ENGINEERING FOR MODELING FIRE CRITICAL INCIDENTS： TOWARDS SUSTAINABLE SECURITY

An intelligent security systems engineering approach is used to analyze fire and explosive critical incidents, a growing concern in urban communities. A feed-forward back-propagation neural network models the damages arising from these critical incidents. The overall goal is to promote fire safety and sustainable security. The intelligent security systems engineering prediction model uses a fully connected multilayer neural network and considers a number of factors related to the fire or explosive incident including the type of property affected, the time of day, and the ignition source. The network was trained on a large number of critical incident records reported in Toronto, Canada between 2000 and 2006. Our intelligent security systems engineering approach can help emergency responders by improving critical incident analysis, sustainable security, and fire risk management.     

Evaluating Efficiency of Decision Making Units with Network Structures

This paper considers the problem of evaluating efficiency of Decision Making Units （DMUs） with network structures of divisions by the Data Envelopment Analysis （DEA） model. All divisions in the network are under a decentralized authority organiza- tion. That is, each division in a decision making unit has its own authority to adjust its input and output. By incorporating the division operations in the DEA model, we discuss the sufficient and necessary conditions for a DMU to be network efficient in series structure and general structure respectively.     

性能优化的认知无线电网络安全路由选择算法

针对认知无线电网络可用频谱动态变化的特点和认知节点存在有意或无意丢包干扰问题,将跳数最少、鲁棒性强和路由安全等级高作为路由选择标准,提出了一种路由选择算法。跳数最少可简化数据传输过程;鲁棒性强可保证在可用频谱动态变化时,数据稳定传输,提高网络吞吐率;路由安全等级高可以减少由于转发节点存在的有意或无意干扰而引起的数据包丢失。给出了算法的具体实现步骤,理论分析和仿真实验表明该算法是高效和可行的。     

最优线性网络编码的分布式构造方法

针对单源组播网络,在合理的假设条件下,提出了最优线性网络编码(在吞吐率最大的前提下编码代价最小)的分布式构造方法。首先对线性网络编码的机理进行分析,得出了不同组播率下编码方案间的内在联系,导出了一个在线测试组播容量的策略;提出的方法以遗传算法为搜索引擎,结合随机线性网络编码,嵌入了在线测试组播容量的策略。通过合理地设计编码系数的生成规则和进化策略,在获得最大吞吐率前提下达到最小编码信道数的同时,构造出相应的编码方案,且编码系数保存在各节点中,从而可以采用确定性网络编码数据传输策略传输数据。仿真结果表明提出的方法是有效的。     

模糊粗糙神经网络的结构与参数优化

基于模糊粗糙隶属函数,建立了一种五层结构的模糊粗糙神经网络（fuzzy rough neural network, FRNN）,对神经元之间的连接,引入一个开关函数,从而把结构优化和参数学习问题转化为单纯的函数优化问题。提出一种混合智能优化算法（hybrid intelligent optimization algorithm, HIOA）用于FRNN的结构和参数优化,适应度函数同时考虑模型的精确性和网络的节俭性。典型的实验结果表明,FRNN适用非线性系统建模,相对于普通神经网络及其优化方法能获得更高的精度和泛化能力。     

Design of complex FIR filters with arbitrary magnitude and group delay responses

To design approximately linear-phase complex coefficient finite impulse response (FIR) digital filters with arbitrary magnitude and group delay responses, a novel neural network approach is studied. The approach is based on a batch back-propagation neural network algorithm by directly minimizing the real magnitude error and phase error from the linear-phase to obtain the filter’s coefficients. The approach can deal with both the real and complex coefficient FIR digital filters design problems. The main advantage of the proposed design method is the significant reduction in the group delay error. The effectiveness of the proposed method is illustrated with two optimal design examples.     

Novel encryption model for multimedia data

A novel encryption model is proposed. It combines encryption process with compression process, and realizes compression and encryption at the same time. The model’s feasibility and security are analyzed in detail. And the relationship between its security and compression ratio is also analyzed.