基于动态主元分析法的传感器故障检测

提出了一种基于动态主元分析的传感器故障检测方法.利用数据矩阵前t时刻和当前时刻的数据,建立多变量多时刻的自回归统计模型.计算主元数据矩阵,建立动态主元模型.以测量速度最慢的传感器的测量周期为统一采样周期,4个连续采样周期为一个诊断周期,建立动态三维测量矩阵,采用残差的平方预报误差的指数加权移动平均(Squared prediction error-Exponentially weighted moving average,SPE-EWMA)模型检测传感器故障.在只存在传感器故障的前提下,模拟发动机开车过程中几种典型的渐变性故障和突变性故障,实验结果表明,算法实时跟踪了各种检测指标的变化,准确检测出故障传感器.     

基于动态跟踪的CPLD振动数据采集设计

提出了一种基于PLL的采样周期动态跟踪的整周期采样设计方案,采用PLL与CPLD技术设计了锁相环倍频电路,实现了采样周期动态跟踪与整周期采样点数滑动可调.在旋转机械轴振动信号监测系统应用中明显地降低了信号采样中频谱泄漏.满足高速整周期采样要求.     

基于动态可行域划分的SVM主动学习

针对传统SVM主动学习中批量采样方法的不足,提出了动态可行域划分算法.从特征空间与参数空间的对偶关系入手,深入分析SVM主动学习的本质,将特征空间中对样本的标注视为参数空间中对可行域的划分;通过综合利用当前分类模型和先前标注样本两方面信息,动态地优化可行域划分方案,以确保选取的样本对模型改进的价值,最终实现更为高效的选择性采样.实验结果表明,基于动态可行域划分的SVM主动学习算法能够显著提高所选样本的信息量,从而能够在有限的标注代价下大幅提高其分类性能.     

利用动态二进制翻译加速应用程序行为特征分析

应用程序运行时典型行为特征分析的一种重要方法是SimPoint,但是为SimPoint生成基本块向量剖析（basic block vector profile,BBV profile）文件非常耗时.首先提出了一个利用动态二进制翻译技术生成BBVprofile的通用框架DBT-BBV,然后详细分析了几种降低开销的优化技术,最后基于DBT-BBV和提出的优化技术设计实现了一个高效的BBVProfile收集工具QPoint.利用SPEC2006测试程序集评估了所提出的优化技术和QPoint的性能和开销.与现有工具相比,QPoint有两个优势:①QPoint的性能高于现有工具,在普通PC机上最高速度为292MIPS,平均速度为109MIPS,BBV Profile收集的平均开销小于4%,在同类工具中最低;②QPoint支持众多体系结构平台,包括x86/x86₆4,ARM,POWER,SPARC,MIPS等,并且可跨指令集收集BBVProfile.结果显示,动态二进制翻译技术在应用程序行为特征分析加速方面具有非常好的效果.     

面向主动学习的模糊核聚类采样算法

针对主动学习中构造初始分类器难以选取代表性样本的问题,提出一种模糊核聚类采样算法。该算法首先通过聚类分析技术将样本集划分,然后分别在类簇中心和类簇边界区域选取样本进行标注,最后依此构造初始分类器。在该算法中,通过高斯核函数把原始样本空间中的点非线性变换到高维特征空间,以达到线性可聚的目的,并引入了一种基于局部密度的初始聚类中心选择方法,从而改善聚类效果。为了提高采样质量,结合划分后各类簇的样本个数设计了一种采样比例分配策略。同时,在采样结束阶段设计了一种后补采样策略,以确保采样个数达标。实验结果分析表明,所提算法可以有效地减少构造初始分类器所需的人工标注负担,并取得较高的分类正确率。     

程序的动态完整性:模型和方法

在信息安全和可信计算中,程序的动态完整性是一个重要问题,特别是无线传感器网络、云计算平台等这一类开放松耦合环境下,怎样度量程序行为的动态完整性的问题尤为突出.基于硬件的可信计算技术和代码证实技术等都没有解决具体行为的动态性度量这个问题,部分原因是缺少一个动态完整性的模型和相应的理论.针对上述问题,在分析了对程序动态完整性安全的威胁基础之上,提出了一个基于密码学的动态完整性理论模型,该模型刻画了程序动态完整性安全的各个要素.基于该模型提出了编译器辅助的流嵌入法,给出了此方法的示例,并分析了此方法的安全性和效率,最后讨论了编译器支持的相关问题,解决了动态完整性的理论和方法中的部分问题.     

基于分层欠采样和Bi-GRU的恶意行为检测模型

为解决类不平衡和样本重叠问题,提出一种基于分层欠采样和Bi-GRU的恶意行为检测模型(SSU-BG)。数据预处理及特征模型的构建;基于欧氏距离算法统计出高频样本的最高密度点及类内平均距离,根据样本分布特点将高频类样本依次划分为稀疏区、稀疏区边界区及稠密区3个区域,根据抽取出样本标签数在稠密区内划分出不同层次圆环,计算每个类的不均衡度,计算其均值作为整个样本的采样比例,按照此时的采样比例在稀疏区边界区的圆环域和稠密区进行分层随机欠采样;将文本向量输入训练好的Bi-GRU模型中。实验结果表明,该模型改善了整体检测效果,提高了恶意评论检测率。     

基于聚类和随机参考分类器的对海雷达目标识别算法*

为了进一步提高复杂干扰环境下对海雷达目标识别的泛化能力,提出基于k-medoids聚类和随机参考分类器(RRC)的动态选择集成算法(KMRRC).主要利用重采样技术生成多个基分类器,然后基于成对多样性度量准则将基分类器划分为多个簇,并基于校验数据集为每个基分类器构建相应的RRC模型,最后利用RRC从各个簇中动态选择竞争力最强的部分基分类器进行集成决策.通过寻优实验确定KMRRC的参数设置,随后利用Java调用Weka API在自建的目标全极化高分辨距离像(HRRP)样本库及17个UCI数据集上进行KMRRC与常用的9种集成算法和基分类算法的对比实验,并进一步研究多样性度量方法的选取对KMRRC性能的影响.实验验证文中算法在对海雷达目标识别领域的有效性.     

一种基于随机采样的SPM管理机制

嵌入式系统对于功耗和面积具有很高的要求.便签存储器(scratchpad memory,SPM)与同等容量Cache相比具有能耗低、片上面积小等优点,现已成为嵌入式处理器中广泛采用的片上存储器.高效的SPM管理策略对于降低系统功耗具有重要意义.传统的SPM管理策略通过编译器采用软件方式进行.随着移动设备及网络互联设备的发展,嵌入式程序的部署方式已趋于多样化,致使传统基于程序特征分析(profiling)的SPM管理方式在某些方面存在局限.提出了一种软硬件结合的基于随机采样(random sampling)的动态SPM管理策略,通过实时监控程序访存特征等手段在运行时动态预测核心工作集(core working set).该方法区别于传统方法之处在于无需依赖profiling信息和编译器进行SPM管理,而通过跟踪程序运行时访存动态特征指导SPM管理.实验表明,该方法可以充分发挥SPM在功耗、面积等方面的优势;通过与一种经典的SPM管理策略相比,所提出的方法在保证系统性能不降低的前提下,提高了SPM管理的灵活性、通用性.     

Query optimization via contention space partitioning and cost error controlling for dynamic multidatabase systems

A multidatabase system (MDBS) integrates information from multiple autonomous local databases. Performing global query optimization to achieve efficient query processing in such a system is challenging due to local autonomy of the data sources. Dynamic factors in the environment make the problem even more difficult. In this paper, we present two techniques, i.e., contention space partitioning and cost error controlling, to perform global query optimization in a dynamic MDBS. Both techniques generate an execution plan with multiple versions for a query in a dynamic MDBS, utilizing the multistate cost models built for the dynamic environment via our previous multistate query sampling method. The first technique partitions the contention space of a dynamic multidatabase environment into a given number of subspaces and chooses a good query execution plan version for each subspace, while the second technique selects a set of execution plan versions by using a given error tolerance to control query execution costs. Experiments demonstrate that the proposed techniques are quite promising for performing global query optimization in a dynamic MDBS. Compared with related work on dynamic query optimization, our approach has an advantage of avoiding the high overhead for modifying or re-generating an execution plan for a query based on dynamic runtime information. Research was supported by the US National Science Foundation under Grant # IIS-9811980 and The University of Michigan.     

Customized Dynamic Load Balancing for a Network of Workstations

Load balancing involves assigning to each processor work proportional to its performance, thereby minimizing the execution time of a program. Although static load balancing can solve many problems (e.g., those caused by processor heterogeneity and nonuniform loops) for most regular applications, the transient external load due to multiple users on a network of workstations necessitates a dynamic approach to load balancing. In this paper we show that different load balancing schemes are best for different applications under varying program and system parameters. Therefore, application-driven customized dynamic load balancing becomes essential for good performance. We present a hybrid compile-time and run-time modeling and decision process which selects (customizes) the best scheme, along with automatic generation of parallel code with calls to a run-time library for load balancing.     

Metadata for Managing Grid Resources in Data Mining Applications

The Grid is an infrastructure for resource sharing and coordinated use of those resources in dynamic heterogeneous distributed environments. The effective use of a Grid requires the definition of metadata for managing the heterogeneity of involved resources that include computers, data, network facilities, and software tools provided by different organizations. Metadata management becomes a key issue when complex applications, such as data-intensive simulations and data mining applications, are executed on a Grid. This paper discusses metadata models for heterogeneous resource management in Grid-based data mining applications. In particular, it discusses how resources are represented and managed in the Knowledge Grid, a framework for Grid-enabled distributed data mining. The paper illustrates how XML-based metadata is used to describe data mining tools, data sources, mining models, and execution plans, and how metadata is used for the design and execution of distributed knowledge discovery applications on Grids.     

一种基于历史信息的自适应动态网格作业调度方法

目前,国内外围绕着网格中的作业调度算法已做了大量研究,先后提出了很多调度算法.但是,这些算法并不能很好地适应网格的动态性、自治性和分布性等特征.对此,提出了一种动态的网格作业调度方法-基于历史信息的自适应动态网格作业调度方法ASHI.该方法利用每个资源上最近作业的执行信息自适应调整预测模型,然后再根据网格的动态性和实时性等因素,对资源进行反馈选择后将作业提交负载较轻的资源上执行.实验证明,ASHI不但能及时有效地对作业进行调度,而且还可有效提高整个网格的吞吐量和均衡系统的负载.     

一个基于元数据导航的服务工作流装配模型

以城市应急处置业务作为应用背景,提出了一个工作流的分层概念模型和一个与之关联的元数据分层描述规范,在此基础上构建了一个基于元数据导航的、从高层的业务应用到底层的分布、动态资源的逐层绑定的运行机制,基于该机制实现了一个城市应急联动系统IERS(Integrated Emergency Response System)的实验原型．该机制解决了从应急处置业务工作流到底层分布服务和资源的装配问题,增强了应急系统对动态、分布服务环境的支持力度,提高了应急处置业务在执行过程中的自动化水平和自适应性,同时也分解和简化了工作流问题的复杂度．     

Minimum effort factor approach for sensor-based control of a four-joint 6-dof redundant manipulator

A sensor-driven control model and a minimum effort control algorithm in terms of time and energy expended during the execution of a movement strategy are described and validated for a multijointed cooperating robotic manipulator. Considering smooth, human-like (anthropomorphic) movements, using joint motion profiles achievable in real time as well as sensory information from all joints, and evaluating the total work expended by each manipulator joint during the execution of a movement strategy, a minimum effort motion trajectory is synthesized to precisely and efficiently position the robotic arm end-effector. This sensor-based approach significantly reduces the computational requirements for such cooperative motion. The minimum effort control algorithm generates several human-like arm movement strategies and selects the best strategy on the basis of expendable effort. The algorithm has an inherent basis to deal with obstacles in an efficient way. Detailed examples are described from the simulation studies. © 1994 John Wiley & Sons, Inc.     

Web metadata extraction and semantic indexing for learning objects extraction

Secondary-school teachers are in constant need of finding relevant digital resources to support specific didactic goals. Unfortunately, generic search engines do not allow them to identify learning objects among semi-structured candidate educational resources, much less retrieve them by teaching goals. This article describes a multi-strategy approach for semantically guided extraction, indexing and search of educational metadata; it combines machine learning, concept analysis, and corpus-based natural language processing techniques. The overall model was validated by comparing extracted metadata against standard search methods and heuristic-based techniques for Classification Accuracy and Metadata Quality (as evaluated by actual teachers), yielding promising results and showing that this semantically guided metadata extraction can effectively enhance access and use of educational digital material.     

一种面向多用户的负载感知动态服务选择模型

在面向多用户的动态环境中进行基于QoS 的服务选择需要面临诸多挑战,而动态的服务负载就是其中之一.当前的服务选择方法难以在多用户多业务的开放环境下应对服务执行时的负载动态变化,缺乏实时感知负载的应变能力.针对这一问题,首先,提出一种基于负载等级的服务多维QoS 模型（load level based multidimensional QoS,简称LLBMQoS）;在此基础上,提出了一种面向多用户的负载感知的动态服务选择模型（load-aware dynamic serviceselection model,简称LADSSM）以实现动态负载环境下的服务优化选择.该模型采用两阶段服务选择：在组合服务规划阶段,生成候选服务队列;在组合服务执行阶段,依据当前负载状态实现服务的动态选择;最后,仿真实验的结果表明：该模型较好地适应了多用户动态环境下的服务负载变化,能够在保证用户端到端QoS需求的前提下,及时而有效地提供效用优化的服务选择方案.     

一种高效的基于采样的关联规则挖掘算法

在事务数据集中发现项目间的关联规则是数据挖掘的一个经典问题,但传统的关联规则挖掘方法对于大事务数据集而言,执行效率相对较低。已经有研究表明,采样技术能有效地改善挖掘效率。在分析现有采样方法的基础上,提出了一种新的基于采样的高效关联规则挖掘算法ESMA。该算法采用了更加有效的双向采样策略。通过实验分析表明,该算法明显地加快了大事务数据库中采样的速度,从而降低了CPU时间,而且具有很好的可扩展性。