Convergence of Greedy Algorithms in Banach Spaces

We study the convergence of greedy algorithms in Banach spaces. We construct an example of a smooth Banach space, where the X-greedy algorithm converges not for all dictionaries and initial vectors. We also study the R-greedy algorithm, which, along with the X-greedy algorithm, is a generalization of the simple greedy algorithm in Hilbert space. We prove its convergence for a certain class of Banach spaces. In particular, this class contains, the spaces ^p,p 2.     

兆瓦级NBI加热系统弧流电源故障诊断方法探究

针对中国环流器2号A(HL-2A)装置弧流电源的特点,以及弧流电源故障诊断的条件与现状,提出了结合中性束注入(NBI)控制系统的相关功能模块与基于Matlab/Simulink仿真技术构建的故障字典,以实现弧流电源运行状态的在线监测及其故障模式的诊断。NBI控制系统在监测到弧流电源故障状态后,调用存储的弧流电源输出信号进行分析,提取信号的特征值,然后通过查找弧流电源故障字典以诊断出可能的故障模式。结合实验数据论证了该故障诊断法对弧流电源简单故障定位准确,对复杂故障能缩小定位范围,具有原理简单、可操作性强等优点。     

基于字典学习的稠密光场重建算法

相机阵列是获取空间中目标光场信息的重要手段,采用大规模密集相机阵列获取高角度分辨率光场的方法增加了采样难度和设备成本,同时产生的大量数据的同步和传输需求也限制了光场采样规模.为了实现稀疏光场采样的稠密重建,本文基于稀疏光场数据,分析同一场景多视角图像的空间、角度信息的关联性和冗余性,建立有效的光场字典学习和稀疏编码数学模型,并根据稀疏编码元素间的约束关系,建立虚拟角度图像稀疏编码恢复模型,提出变换域稀疏编码恢复方法,并结合多场景稠密重建实验,验证提出方法的有效性.实验结果表明,本文方法能够对场景中的遮挡、阴影以及复杂的光影变化信息进行高质量恢复,可以用于复杂场景的稀疏光场稠密重建.本研究实现了线性采集稀疏光场的稠密重建,未来将针对非线性采集稀疏光场的稠密重建进行研究,以推进光场成像在实际工程中的应用.     

On Zipf’s law and rank distributions in linguistics and semiotics

A number of formulas of linguistic statistics are refined. The notions of real and virtual cardinality of a sign are introduced. We show that a formula refining Zipf’s law for the occurrence frequencies in frequency dictionaries can be extended to arbitrary sign objects, i.e., semiotic systems.     

An Efficient Dictionary Organization for Maximum Diagnosis

The major problem of fault diagnosis with a fault dictionary is the enormous amount of data. The technique used to manage this data can have a significant effect on the outcome of the fault diagnosis procedure. If information is removed from a fault dictionary in order to reduce the size of the dictionary, its ability to diagnose stuck-at faults and unmodeled faults may be severely debased. Therefore, we focus on methods for producing a dictionary that is both small and lossless-compacted. We propose an efficient dictionary for maximum diagnosis, which is called SD-Dictionary. This dictionary consists of a static sub-dictionary and a dynamic sub-dictionary in order to make a smaller dictionary while maintaining the critical information needed for the diagnostic ability. Experimental results on ISCAS’ 85, ISCAS’ 89 and ITC’ 99 benchmark circuits show that the size of the proposed dictionary is substantially reduced, while the dictionary retains most or all of the diagnostic capability of the full dictionary. This work was supported by the “System IC 2010” project of Korea Ministry of Science and Technology and Ministry of Commerce, Industry and Energy. Editor: Y. Takamatsu Sunghoon Chun received the B.S. degrees in Electrical and Electronic Engineering from Yonsei University, Seoul, Korea, in 2002. He was a Reseach Engineer with ASIC Research Center in Yonsei University. He researched for test methodologies for SoC. He received the M.S. degrees in Electrical and Electronic Engineering from Yonsei University in 2005. He is currently working toward Ph.D. degree in Electrical and Electronic Engineering at Yonsei University. His area of interests includes SoC testing, delay testing, fault diagnosis, functional testing for processor based system and test methodologies for signal integrity faults. Sangwook Kim received the B.S., and M.S. degrees in Electrical and Electronic Engineering from Yonsei University, Seoul, Korea, in 1999, and 2001, respectively. He researched for Digital Signal Processor design and fault diagnosis of VLSI. He is a Research Engineer with SoC Design Group of System IC Division in LG Electronics, Inc. He is currently interested in SoC design for HDTV and design verification. Hong-Sik Kim was born in Seoul, Korea, on April 4, 1973. He received the B.S., M.S. and Ph.D. degrees in Electrical and Electronic Engineering from Yonsei University, Seoul, Korea, in 1977, 1999, and 2004, respectively. He was a Post-Doctorial Fellow with the Institute of Virginia Technology. He is currently working on System LSI Group in the Samsung Electronics. His current research interest includes design-for-testability, built-in self tests and fault diagnosis. Sungho Kang received the B.S. degree from Seoul National University, Seoul, Korea, and the M.S. and Ph.D. degrees in electrical and computer engineering from The University of Texas at Austin. He was a Post-Doctorial Fellow with the University of Texas at Austin, a Research Scientist with the Schlumberger Laboratory for Computer Science, Schlumberger Inc., and a Senior Staff Engineer with the Semiconductor Systems Design Technology, Motorola Inc. Since 1994, he has been an Associate Professor with the Department of Electrical and Electronic Engineering, Yonsei University, Seoul, Korea. His current research interests include VLSI design, VLSI CAD and VLSI testing and design for testability.     

模拟电路故障诊断的新故障字典法

基于节点电压灵敏度,将文献[1]中的线性无容差电路的故障字典法推广到可以诊断容差模拟电路和非线性电路软故障的新故障字典法。讨论了该方法的原理和字典的建立方法,给出了仿真实例。     

基于字典学习方法的CT不完全投影图像重建算法

对于不完全投影角度的重建研究是CT图像重建中一个重要的问题.将压缩感知中字典学习的方法与CT重建算法ART迭代算法相结合.字典学习方法中字典更新采用K-SVD(K-奇异值分解)算法,稀疏编码采用OMP(正交匹配追踪)算法.最后通过对标准Head头部模型进行仿真实验,验证了字典学习方法在CT图像重建中对于提高图像的重建质量和提高信噪比的可行性与有效性.另外还研究了字典学习中图像块大小和滑动距离对重建图像的影响     

基于SAT的组合电路自动测试向量生成

实现了基于可满足性(SAT)求解的方法,以解决固定型和时延故障的自动测试向量生成问题.详细讨论了如何利用电路的拓扑结构以及从ATPG到合取范式(CNF)的编码方法.CNF被输入到一个高效的SAT求解器zchaff中求解.在ISCAS85测试实例中验证了该算法的有效性.     

基于核Fisher判别字典学习的稀疏表示分类

在基于稀疏表示分类的模式识别中,字典学习(DL) 可以为稀疏表示获得更为精简的数据表示。最近的基于Fisher判别的字典学习(FDDL)可以学 习到更加判别的稀疏字典,使得稀疏表示分类具有很强的识别性能。核空间变换可以学习到 非线性结构信息,这对判别分类非常有用。为了充分利用 核空间特性以学习更加判别的稀疏字典来提升最终的识别性能,在FDDL的基础上,提出了两 种核化的稀疏表示DL方法。首先原始训练数据被投影到高维核空间,进行基于Fisher 判别的核稀 疏表示DLFDKDL;其次在稀疏系数上附加核Fisher约束,进行基于核Fisher判别的核稀疏表 示DL(KFDKDL),使得所学习的字典具有更强的判别能力。在多个公开的图像数据库上的稀疏 表示分类实验结果验证了所提出的FDKDL和KFDKDL方法的有效性。     

基于稀疏采样与级联字典的微波辐射图像重构方法*

针对微波辐射测量成像系统在一次观测中所采集的数据量大,基于奈奎斯特采样的常规,微波辐射成像方法难以实现高分辨率要求。本文在挖掘微波辐射图像多结构信息可压缩的基础上,利用随机观测矩阵对微波辐射图像进行线性压缩投影,减少数据采集量,降低系统的复杂性。考虑微波辐射图像具有多结构形式,采用单一的正交基难以稀疏表示复杂场景的微波辐射图像,利用全变差分和小波的级联字典对微波辐射图像进行稀疏表示,然后用OMP 算法重构微波辐射图像。仿真实验结果表明:级联字典重构微波辐射图像的性能优于单一正交基。