Bayes网络与关系模型的集成:概率关系模型

对象关系模型和Bayes网络分别是关系理论和概率理论两个不同领域中最重要的模型，它们首次集成于本文引入的概率关系模型中，作为新型的概率模型，概率关系模型不仅继承了Bayes网络的大部分优点，而且关系特征和对象的概念使它能有效地克服Bayes网络在许多方面的不足，而成为对复杂系统模建的理想工具，是对Bayes网络的重要创新；作为新型的关系模型，概率关系模型也是对传统关系模型的重要创新，具备概率特征的对象关系模型有了处理不确定性问题的能力。概率关系模型的创建对复杂智能信息系统开发研究有有着特别重要的意义，本文首先评述Bayes网络和对象关系模型，然后在此基础上引入概率关系模型。     

超声红外热像检测条件的优化方法研究

在超声红外热像检测中,激励强度、激励时间和工具杆与被测对象之间的预紧力等检测条件是影响缺陷热信号的重要因素。实验和仿真分析表明:激励时间和激励强度的增加将使裂纹生热增强,而预紧力的增加却使得裂纹生热呈现先增强后减弱的趋势,且检测条件对裂纹生热存在交互影响;另外,不恰当的检测条件将导致超声激励系统报警。基于检测条件对裂纹热信号和报警数据的影响分析,文中提出了采用多元非线性回归模型和Logistic回归模型以估算特定检测条件下裂纹检出概率和检测报警概率,最终确定了检测条件的选择范围。上述检测条件的优化方法能够为超声红外热像检测技术中检测方案的制定提供理论指导。     

一种条件概率与遗传算法相结合的分类方法

数据挖掘就是从大量数据中发现以前未知的有用信息、模式、趋势的过程。分类是数据挖掘的一种主要方法。文章指出分类的实质是找出各属性对分类的贡献大小，然后采用分而治之的思想，先用条件概率的方法计算单个属性对分类的贡献，再利用遗传算法计算各属性对分类的重要程度，提出了条件概率与遗传算法相结合的分类方法，利用UCI数据集进行验证，并与相同条件下的其它分类方法进行了比较，实验表明该方法是一种简单有效的分类方法。     

一种类图转化为关系数据库的自适应算法研究

提出了一种将类图转化为关系数据库RDBS的算法,首先使用Rational Rose工具创建类图,并使用Java语言对其进行描述;其次,将源代码文本作为算法执行,通过特定算法比较“ClaSS”,“attribute”和“relation”等关键字,区分出类、属性和关系;在完成数据库表的名称、属性字段的类型及大小等设置描述后,在MYSQL数据库中执行该算法,生成SQL命令创建数据库和表。这种算法能减少类图与关系数据库之间转化的出错概率和时间。     

概率图模型在无线网络推理中的应用

重点研究概率图模型发展和分类,其信号处理方法可以有效解决无线网络中一些不确定性推理问题。在无线理论研究中,讨论并分析因子图的一些已有成果,同时关注到因子图方法在无线传感网络中的应用。结合无线网络的关键技术要求对该类模型的应用领域做进一步探讨,展望未来图模型在该领域中的应用前景。     

概率计算在通信信号处理系统实现中的应用

概率计算方法是一种数值表征和计算方法;在概率计算法中,传统的基本数值计算可以与数字电路中的基本逻辑门电路建立直接的映射关系。针对现代无线通信系统中数字信号处理单元面临的复杂度、功耗和运算速度等挑战,结合通信信号处理中所呈现的统计特性,概率计算方法为通信系统数字信号处理实现提供了一条新颖的技术路线。在讨论、分析了该技术在通信信号处理系统集成电路实现中应用的可行性和应用桥梁的基础上,介绍了概率计算在信号处理系统中的典型应用,最后分析了该技术在通信信号处理系统中的挑战。     

贝叶斯网络在远程教学中的应用

贝叶斯网目前广泛应用于专家系统中,用于处理大量以条件概率为形式的数据.首先对贝叶斯网络进行概述,论述其在远程教学学生模型中的应用.然后重点介绍学生模型和学生评估模型的结构、功能和概率推理算法.模拟以贝叶斯网为学生模型的远程教学系统.实践表明,该系统能较真实地反映学生当前学习问题.     

反导跟踪雷达最优搜索时序研究

提出了一种基于概率轨迹的相控阵雷达最优波位搜索时序模型.该模型根据预警信息的统计特性,将目标的位置和速度引导误差分布等效为多条目标以一定概率沿其飞行的航迹,即概率轨迹.通过递推的方式实时计算更新各轨迹的概率,从而动态地得到各波位的目标落入概率,依据最大累积发现概率准则确定雷达波位的搜索时序.仿真结果充分验证了该模型在高速运动目标条件下具有比传统的最优搜索模型更佳的探测性能.     

分组样本下先验BN模型及条件概率的学习算法

文章扩展经典的先验BN模型，采用两层学习结构讨论分组样本下BN模型的条件概率及学习算法：一层是对各组私有条件概率分布的学习；另一层是对各组公有条件概率分布的学习。算法在综合公有后验条件概率分布和本组学习实例数据分布特征的基础上，实现对各组私有条件概率分布的学习，并可通过经验或学习来改变综合值中共性和个性的比例。     

复合制导空空导弹截获概率模型仿真

针对复合制导空空导弹截获目标时导引头预置天线角度与导弹允许截获不在同一时刻这一问题,建立了一种新的截获概率估算模型。根据中末制导交接班时导弹截获目标的算法流程,确立了导弹截获目标的条件,分析了影响导弹截获概率的主要误差源,建立了各误差数学模型,并通过等效转化法将弹目指示位置向量在三维空间的散布转化为弹目实际相对位置向量在三维空间的散布,完成截获概率估算模型的建立。该模型利用已知的误差先验估计及导弹飞行过程中的测量信息,可在一次弹道计算中解算导弹截获概率。利用此模型采用控制变量法分析了各误差因素与截获概率的关系,仿真结果表明：目标截获概率对数据链周期和测角精度敏感性最高。     

低电压下CMOS数字电路概率模型研究

为解决数字电路低功耗问题,电路工作电压被不断降低,导致电路逻辑器件呈现概率特性。本文提出了低电压下CMOS数字电路的错误概率模型,并完成硬件电路测试验证。本文首先详述了深亚微米（DSM）量级的门电路及模块在低电压供电条件下导致器件出错的因素,结合概率器件结构模型推导基本逻辑门概率模型,并提出了状态转移法用于完成由门级到模块级的概率分析模型;我们搭建硬件平台对CMOS逻辑芯片进行了低供电压测试,通过分析理论推导结果与实测结果,验证并完善了分析模型。实验结果表明,由状态转移法推导的电路概率模型符合电路实际性能,从而为构建低电压下数字电路概率模型提供了可靠分析模型。      

概率XML数据模型的综述

随着XML成为网络信息表示和交换的标准以及不确定数据的广泛存在,概率XML数据库管理技术成为了当今研究的热点,研究者根据概率数据的类型和解决实际问题的需要提出了多种概率XML数据模型。首先介绍了概率XML数据管理技术的概念,特点和挑战;其次综述了概率XML数据和概率XML数据模型,各种模型的核心思想都来自于可能世界模型,通过选择孩子节点以及删除节点可得到一个可能世界的实例,而且所有实例的概率之和为1;最后介绍了不同模型之间的转换关系。为概率XML数据库的查询、规范化理论奠定了基础。     

分布式电源的概率建模及其对电力系统的影响

本文对分布式电源的概率建模及其对电力系统的影响进行了简单的分析,从而降低建立生产成本与供应成本,实现资源优化配置,从而促进电力企业快速发展.     

计算通信网络整体概率连通性的一种新算法

本文提出了一种计算网络整体概率连通性的新算法。该算法首先构造了一种多级状态空间分解法,对网络状态空间进行了分解,并对分解所得的生成事件提出了一种提出了一种特殊的生成子网构造法,从而有效减少了网络状态空间集。新算法通过迭代运算用上下界来逼近网络概率连通性的值,与传统算法相比,具有运算速度快、收敛性能好,很适合近似计算的特点。     

Wire congestion estimation in a bounded-bend model

Interconnect congestion estimation plays an important role in the physical design of integrated circuits. Fast congestion analysis prior to global routing enhances the placement quality and improves the routability for the subsequent routing phases. This article presents a novel congestion estimation method for a wire layout with bounded detours and bends. Experimental results on benchmarks demonstrate the efficiency and accuracy of our approach.     

隐马尔可夫模型演化下的隐组查询

针对社会网络图中的隐组查询问题,提出了一种基于隐马尔科夫模型演化的方法。不同于传统方法,文中首先对“微观法则”提出了一些合理的假设,这些法则决定了在某时刻一个个体是否存在于一个特定群体。通过这些假设,可以得到社会个体和群体的动态演化。然后根据群体演化,找出长时间保持通信的群体作为潜在的隐组,再通过进一步分析,确保这些潜在的隐组以一个较高的概率成为理想的结果。为验证算法的有效性,文中分别对模拟和真实的数据进行了测试。     

Parametric Probabilistic Routing in Sensor Networks

Motivated by realistic sensor network scenarios that have mis-in-formed nodes and variable network topologies, we propose an approach to routing that combines the best features of limited-flooding and information-sensitive path-finding protocols into a reliable, low-power method that can make delivery guarantees independent of parameter values or information noise levels. We introduce Parametric Probabilistic Sensor Network Routing Protocols, a family of light-weight and robust multi-path routing protocols for sensor networks in which an intermediate sensor decides to forward a message with a probability that depends on various parameters, such as the distance of the sensor to the destination, the distance of the source sensor to the destination, or the number of hops a packet has already traveled. We propose two protocol variants of this family and compare the new methods to other probabilistic and deterministic protocols, namely constant-probability gossiping, uncontrolled flooding, random wandering, shortest path routing (and a variation), and a load-spreading shortest-path protocol inspired by (Servetto and Barrenechea, 2002). We consider sensor networks where a sensor’s knowledge of the local or global information is uncertain (parametrically noised) due to sensor mobility, and investigate the trade-off between robustness of the protocol as measured by quality of service (in particular, successful delivery rate and delivery lag) and use of resources (total network load). Our results for networks with randomly placed nodes and realistic urban networks with varying density show that the multi-path protocols are less sensitive to misinformation, and suggest that in the presence of noisy data, a limited flooding strategy will actually perform better and use fewer resources than an attempted single-path routing strategy, with the Parametric Probabilistic Sensor Network Routing Protocols outperforming other protocols. Our results also suggest that protocols using network information perform better than protocols that do not, even in the presence of strong noise. Christopher L. Barrett is leader of the Basic and Applied Simulation Science Group of the Computing and Computational Sciences Division at Los Alamos National Laboratory. His Group is a simulation science and technology (S&T) invention organization of 30 scientists devoted to providing large-scale, high performance methods for systems analysis and simulation-based assisted reasoning. His Group engages in fundamental mathematical, algorithmic, and complex systems analysis research. Current applied research is focused on interdependent simulation and analysis tools for complex, socio-technical systems like transportation, communications, public health and other critical infrastructure areas. His scientific experience is in simulation, scientific computation, algorithm theory and development, system science and control, engineering science, bio-systems analysis, decision science, cognitive human factors, testing and training. His applied science and engineering achievements include, for example, development of large-scale, high performance simulation systems (e.g., Transportation Analysis Simulation System, TRANSIMS) and development of a distributed computing approach for detailed simulation-based study of mobile, packet switched digital communications systems (Self Organizing Stochastic Rebroadcast Relay, SORSRER). He has a M.S. and Ph.D. in Bio-information Systems from California Institute of Technology. He is a decorated Navy veteran having served in both the submarine service and as a pilot. He has been awarded three Distinguished Service Awards from Los Alamos National Laboratory, one from the Alliance for Transportation Research, one from the Royal Institute of Technology, Stockholm, and one from Artificial Life and Robotics, Oita University, Japan. Stephan J. Eidenbenz is a technical staff member in the Basic and Applied Simulation Science group (CCS-5) at Los Alamos National Laboratory (LANL). He received an M.Sc. in Computer Science from the Swiss Federal Institute of Technology (ETH) in Zurich in 1997 and a Ph.D. in Computer Science from ETH in 2000; he also obtained a Bachelor’s degree in business administration from GSBA in Zurich in 1999. Stephan has worked for McKinsey & Co. in Switzerland, where he received training in business administration and microeconomics. He has held a postdoctoral position at ETH and he has been a postdoctoral fellow at LANL. Stephan’s more than 30 publications cover a wide range of subjects such as approximability and inapproximability properties of visibility problems in polygons and terrains, error modeling in sequencing problems for computation biology, and designing communication protocols robust against selfish behavior. His current research interests include selfish networking, algorithmic game theory, network modeling and simulation, network design, and network optimization. Lukas Kroc is a student of M.Sc. program in Computer Science at Charles University in Prague. In 2003, he was a Graduate Research Assistant at the Basic and Applied Simulation Science group (CCS-5) at Los Alamos National Laboratory. His research interests include simulation, wireless networking and artificial intelligence. Madhav V. Marathe is a Team Leader for Mathematics and Computer Science in the Basic and Applied Simulation Science group, Computer and Computational Sciences (CCS-5) at the Los Alamos National Laboratory. He obtained his B.Tech in 1989 in Computer Science and Engg. from IIT Madras, India and his Ph.D. in 1994 in Computer Science, from University at Albany. His team focuses on developing mathematical and computational tools for design and analysis of large scale simulations of socio-technical and critical infrastructure systems. His research interests are in modeling and simulations of large socio-technical systems, design and analysis of algorithms, computational complexity theory, theory of parallel, distributed and mobile computing and communication systems. He has published over 100 research articles in peer reviewed journals and conferences. He is an adjunct faculty in the Computer Science Department at the University of New Mexico. James P. Smith is a technical staff member in the Basic and Applied Simulation Science Group of the Computing and Computational Sciences Division at Los Alamos National Laboratory. His principal interest is in high performance computing applied to modeling, simulation and analysis of socio-technical systems. His current research applies to national infrastructure, especially telecommunication/computing, public health, and transportation. He has scientific experience in high performance computing and parallel processing applied to large-scale microscopic simulations, including original software design and debugging of very large, evolving systems of inter-operable computational systems, and efficient analysis and synthesis of massive data produced by multi-scale complex environments. Before attending graduate school he worked for a short time in nuclear theory, and had several publications in experimental biophysics from the Pennsylvania Muscle Institute and Bockus Research Institute. During graduate school he took a one year hiatus to start a company to work in analytic finance, and then spent time doing theoretical space physics at LANL. His graduate work eventually included theoretical and experimental fusion research, but concentrated on computational space plasma physics. He has publications in biophysics, analytic finance, education, space plasma physics and computer science, and is a co-inventor on the TRANSIMS patent. He has a Ph.D. in Theoretical Plasma Physics from the University of Texas at Austin.This revised version was published online in August 2005 with a corrected cover date.     

一种新的联合概率数据互联算法

点迹与航迹数据互联是多目标跟踪问题中迫切需要解决的问题。分析了目前解决数据互联问题的方法与最新研究成果，建立了一个多目标数据互联模型，提出了一种新的联合概率数据互联算法，实现了微机仿真并给出了仿真结果。     

一种基于变分推断的雷达多目标跟踪JPDA算法

针对雷达邻近多目标跟踪问题,提出了一种基于变分推断的联合概率数据关联算法（Joint Probability Data Association,JPDA）。通过建立关于目标状态和两个关联指示的概率图模型,并根据不同变量之间的信息传递构造对应的自由能目标函数,迭代该目标函数求解出目标和当前检测量测之间的最佳边缘关联概率。将所提算法与经典JPDA和k 近邻联合概率数据关联(k Nearest Neighbor-Joint Probability Data Association,kNN－JPDA) 算法进行对比,结果表明新算法具备更高的跟踪位置精度,并且能够有效地避免因邻近目标数量增多而引起的计算上的组合爆炸问题。     

基于概率策略逻辑的VANET信息广播模型定量验证方法研究

VANET网络中信息的发送和接收具有随机性和不确定性,IEEE 802.11p广播协议无法适应VANET网络拓扑动态变化,于是研究者们根据不同环境中的具体应用需求提出了各种VANET广播协议,如何对新提出的协议的性能以及可靠性进行分析与验证是一个关键性问题.自动化的定量验证技术能够针对系统需要满足的多个性质进行分析,并给出满足需求的最大或者最小概率.然而研究人员在进行定量验证过程中使用的PTCL、rPATL等逻辑语言都不能够明确描述用户的策略是什么,因此本文提出基于概率策略逻辑的模型定量验证方法.该方法首先对系统中的多个角色使用概率时间接口自动机对其行为建模,然后使用概率策略逻辑语言对系统需要满足的性质进行描述,最后基于定量验证算法自动给出系统相关性质的分析结论.本文将该方法应用到VANET信息广播协议性能分析上,能够针对外界环境的变化选择合理的策略,从而分析出不同环境下信息广播发送成功的最大概率.