基于支持向量机的信息系统性能评估

根据信息系统评价原则,提出了信息系统性能评价的指标体系,并对各指标进行了定量描述,后利用支持向量机（SVM）的分类算法,建立了基于SVM的性能评估模型,实现了信息系统评估的自动化。通过测试,证明了该方法的有效性和可行性。     

分布式并行系统中的一种故障分类模型

故障诊断是分布式并行环境下容错系统的关键部分,故障分类模型是影响故障诊断性能的重要因素之一。由于不同的分布式系统有其不同的特点．为了减少系统在故障诊断方面的负担,故障诊断方案一般都考虑程序的需求和系统的属性,选择最合适的故障分类模型。本文提出了一种新的分布式并行环境下的故障分类模型,可以将故障诊断限定在一个合理的故障集中。将这种分类模型和特殊的程序需求以及系统属性相结合,能够得到一个效果理想的故障检测方案。     

基于状态可行域约束的极值搜索系统预设性能控制

针对极值搜索系统在稳定状态存在可行域约束情况下的预设性能控制问题,首先利用罚函数构造新型目标函数并设计极值参考轨迹;然后,通过性能函数和误差转换函数构建系统等效模型,利用反演控制方法进行预设性能控制器设计;最后,设计不确定参数自适应估计律.系统在实现极值搜索的同时能够满足预先设定的性能指标.数值仿真验证了所提出方法的有效性.     

深度学习在肺结节辅助诊断中的应用

肺癌位居癌症死亡率首位;对其进行早期诊断和治疗可降低肺癌患者的死亡率。深度学习能够自动提取结节特征;并完成肺结节的良恶性及恶性等级分类;因此深度学习方法成为肺癌早期诊断的重要手段。对常用数据集进行介绍;系统阐述了栈式去噪自编码器（SDAE）、深度置信网络（DBN）、生成对抗网络（GAN）、卷积神经网络（CNN）、循环神经网络（RNN）和迁移学习技术在肺结节良恶性分类中的应用;阐述了深度卷积生成对抗网络（DCGAN）、多尺度卷积神经网络（MCNN）、U型网络（U-Net）和集成学习技术在肺结节恶性等级分类中的应用;针对肺结节分类的深度学习方法进行了综合分析;并对未来研究方向进行展望。     

基于视频的车辆几何参数提取与车型分类

针对车辆自动分类需求,通过引入地理信息系统空间形态的描述方法,给出车辆侧面图像的几何信息描述参数和计算方法,设计用于车辆分类的几何特征参数及其组合参数,基于侧视轮廓的几何特征,提出车辆轮轴参数的自动提取方法,设计一个基于几何形态参数的车型自动分类器。实验证明该方案的有效性。     

一类复杂经济系统的建模方法与应用

针对我国经济体制转型时期的特殊情况，本文提出一种新的建模方法，并对Ｌ省２０００年的科技人才需求进行了预测。     

Parametric Analysis of Computer Systems

A general parametric analysis problem which allows the use of parameter variables in both the real-time automata and the specifications is proposed and solved. The analysis algorithm is much simpler and can run more efficiently in average cases than can previous works.     

Nudelman interpolation, parametrizations of lossless functions and balanced realizations

We investigate the parametrization issue for discrete-time stable all-pass multivariable systems by means of a Schur algorithm involving a Nudelman interpolation condition. A recursive construction of balanced realizations is associated with it, that possesses a very good numerical behavior. Several atlases of charts or families of local parametrizations are presented and for each atlas a chart selection strategy is proposed. The last one can be viewed as a nice mutual encoding property of lossless functions and turns out to be very efficient. These parametrizations allow for solving optimization problems within the fields of system identification and optimal control.     

线阵CCD探测器主要性能的测试与研究

为了保证 CCD 探测器在航空航天探测方面的可靠性,本文针对Kodak公司的商用三线阵CCD 探测器KLI-2113,给出了该三线阵CCD 探测器的主要测试性能参数及其定义,给出了该三线阵 CCD 探测器的测试原理和测设方法,并且对部分核心设备指标进行了说明。最终对20片KLI-2113进行测试,得到了具有参考意义测试结果,并对实验结果进行了对比分析,测试结果在片内和片间都有较好的一致性,该结果对三线阵探测器的航空航天等遥感观测领域的运用提供了有力的参考数据。     

Concept classifier for knowledge acquisition in frame-based systems

Frame-based systems that employ inheritance networks as a form of knowledge representation have a number of inherent knowledge acquisition problems, one of the most significant being the transfer to the representation system of knowledge itself. The problem of concept classification, and specifically that of determining the location of a new concept in an existing network inheritance hierarchy, is discussed here using an experimental knowledge-base editor, KRE. Tools which support the process of knowledge base construction must allow the user to concentrate on the domain problems, and not on low level, representation system decisions. KRE, written in C, is a knowledge acquisition tool which assists the knowledge engineer by using an interactive acquisition strategy during the process of concept classification. The processes of classification, and its advantages over other knowledge representation systems, are presented.     

基于粗糙集约简的多分类器系统构造方法

多分类器系统是近年来兴起的一种有效的分类机制,为提高多分类器系统的分类精度,提出了一种基于粗糙集约简构造多分类器系统的机制,并从输入和输出两个角度对如何选择单个分类器进行了探讨。通过对4个UCI数据集进行验证,发现基于输出的选择融合方法得到了最好的分类效果。     

网络测量指标初步分析

网络测量指标是进行网络性能分析的基础。介绍了两种国际标准化组织提出的网络测量指标,并进行比较,并给出实际测量时的常用指标     

实时系统中的调度算法分类研究

本文叙述了实时系统中调度算法的分类、各类算法的研究成果和近期研究状况。     

A cost-based fuzzy system for pattern classification with class importance

This paper proposes a cost-based fuzzy classification system for pattern classification problems with an order of class importance. The task here is to minimize the misclassification of patterns from an important class. It is assumed that the classification importance is given for each class, not for each pattern. Another assumption is that only the order of importance is given for given classes without any numerical measures of importance. We show the performance of the proposed cost-based fuzzy classification system for a real-world pattern classification problem. This work was presented in part at the 12th International Symposium on Artificial Life and Robotics, Oita, Japan, January 25–27, 2007     

Genetic synthesis of a fuzzy pulse discriminator in electrical discharge machining

In this paper, a sensor system, called a fuzzy pulse discriminator, is developed to classify various discharge pulses in electrical discharge machining (EDM). The fuzzy rules of the pulse discriminator are obtained based on the features of the gap voltage and gap current between the tool workpiece. The membership functions of the fuzzy pulse discriminator are automatically synthesized by using genetic algorithms. The effectiveness of this approach is verified under different cutting parameters.     

Using eager strategies to improve NFS I/O performance

Typical network file system (NFS) clients write lazily: they leave dirty pages in the page cache and defer writing to the server. This reduces network traffic when applications repeatedly modify the same set of pages. However, this approach can lead to memory pressure, when the number of available pages on the client system is so low that the system must work harder to reclaim dirty pages. We show that NFS performance is poor under memory pressure and present two mechanisms to solve it: eager writeback and eager page laundering. These mechanisms change the client's data management policy from lazy to eager, in which dirty pages are written back proactively, resulting in higher throughput for sequential writes. In addition, we show that NFS servers suffer from out-of-order file operations, which further reduce performance. We introduce request ordering, a server mechanism to process operations, as much as possible, in the order they were sent by the client, which improves read performance substantially. We have implemented these techniques in the Linux operating system. I/O performance is improved, with the most pronounced improvement visible for sequential access to large files. We see 33% improvement in the performance of streaming write workloads and more than triple the performance of streaming read workloads. We evaluate several non-sequential workloads and show that these techniques do not degrade performance, and can sometimes improve performance.     

A MULTIVARIABLE STABILITY MARGIN IN THE PRESENCE OF TIME-VARYING,BOUNDED RATE GAINS

In this paper we consider a MIMO asymptotically stable linear plant. For such a system the classical concepts of quadratic stability margin and multivariable gain margin can be defined. These margins have the following interpretation: consider the closed-loop system composed of the plant and several real parameters, one inserted in each channel of the loop; then any time-varying (time-invariant) parameters whose amplitudes are smaller than the quadratic stability (multivariable gain) margin result in a stable closed-loop system. For time-varying parameters whose magnitudes are between these two stability measures, stability may depend on the rate of variation of the parameters. Therefore it makes sense to consider the stability margin given by the maximal allowable rate of variation of the parameters which guarantees stability of the closed loop system. As shown in this paper, a lower bound on this margin can be obtained with the aid of parameter dependent Lyapunov functions. © 1997 by John Wiley & Sons, Ltd.