An Adaptive Approach to Schema Classification for Data Warehouse Modeling

Data warehouse （DW） modeling is a complicated task, involving both knowledge of business processes and familiarity with operational information systems structure and behavior. Existing DW modeling techniques suffer from the following major drawbacks -- data-driven approach requires high levels of expertise and neglects the requirements of end users, while demand-driven approach lacks enterprise-wide vision and is regardless of existing models of underlying operational systems. In order to make up for those shortcomings, a method of classification of schema elements for DW modeling is proposed in this paper. We first put forward the vector space models for subjects and schema elements, then present an adaptive approach with self-tuning theory to construct context vectors of subjects, and finally classify the source schema elements into different subjects of the DW automatically. Benefited from the result of the schema elements classification, designers can model and construct a DW more easily.     

基于深度学习的分层关联多行人跟踪

由于连续帧之间数据关联的不确定性和所提取外观特征的鉴别力不足,多目标跟踪容易受目标外观变化、运动状态变化、相似目标以及目标消失再出现等干扰因素的影响,出现轨迹ID变换的问题,从而限制基于轨迹分析的行为检测、姿态识别等研究的性能。为了提高数据关联的可靠性从而减少轨迹ID变换,提出了一种基于轨迹置信度的分层数据关联方式。同时,为了提高用于数据关联的外观特征的鉴别力,提出融合了过程层的特征提取网络。在公开的MOT16测试数据集上验证表明,该算法在保证跟踪准确度和精确度的同时,有效降低了轨迹ID变换的次数,提高了跟踪性能。     

Proactive scheduling in distributed computing—A reinforcement learning approach

In distributed computing such as grid computing, online users submit their tasks anytime and anywhere to dynamic resources. Task arrival and execution processes are stochastic. How to adapt to the consequent uncertainties, as well as scheduling overhead and response time, are the main concern in dynamic scheduling. Based on the decision theory, scheduling is formulated as a Markov decision process (MDP). To address this problem, an approach from machine learning is used to learn task arrival and execution patterns online. The proposed algorithm can automatically acquire such knowledge without any aforehand modeling, and proactively allocate tasks on account of the forthcoming tasks and their execution dynamics. Under comparison with four classic algorithms such as Min–Min, Min–Max, Suffrage, and ECT, the proposed algorithm has much less scheduling overhead. The experiments over both synthetic and practical environments reveal that the proposed algorithm outperforms other algorithms in terms of the average response time. The smaller variance of average response time further validates the robustness of our algorithm.     

一种适用于嵌入式环境下的手写汉字分割算法

汉字分割是连续手写汉字识别的关键问题之一,现有多数汉字分割算法需要对笔划结构进行分析,由于汉字具有复杂的结构和不同的书写风格,因此往往计算量偏大.针对嵌入式系统处理能力和内存资源相对较少的特点,介绍了一种基于笔划连通域和字符投影的手写汉字分割算法,以连通域为单元实现连续汉字的分割,不用考虑连通域中所包含笔划的结构和风格,降低了计算量.     

SDN网络虚拟化中规则映射研究

软件定义网络（SDN）为网络虚拟化提供了新的解决方案,通过网络虚拟化技术可以将一套基础设施虚拟化为多个逻辑网络从而满足不同的网络需求.本文研究了SDN网络虚拟化时多个物理交换机虚拟为一个大虚拟交换机的过程中,虚拟网络规则与物理网络规则的映射问题.综合考虑链路负载、规则分布以及节点负载,提出了三段式规则映射优化算法.首先根据虚拟网络的规则请求生成组播源节点和目的节点集,采用MPH算法生成规则映射树;然后采用入节点最近原则,将虚拟网络规则请求的指令序列部署到规则映射树中的中间节点和叶子节点中;最后考虑节点负载,对规则部署进行微调,最终生成虚拟规则映射策略.通过仿真实验,与直接边缘节点部署相比,平均降低了网络节点规则总数量40%以上.     

Multimode resonator based on composite right‐/left‐handed transmission line for UWB bandpass filter application

A new multimode resonator (MMR) using composite right‐/left‐handed transmission line (CRLH TL) is proposed and discussed. The CRLH TL structure is constructed by cascading interdigital coupled microstrip line sections on which short‐ended stepped impedance stubs are loaded. Dispersion characteristic of the transmission line structure is obtained using the Bloch–Floquet theory. The resonator, which has multiresonances electrical behaviors, is especially suitable in ultrawideband (UWB) applications. An UWB filter is presented as an illustration. With transmission zeros introduced at upper stopband, the filter has a sharp skirt performance. In addition, rejection level at lower stopband also gets enhanced due to direct current suppression effects of the multimode resonator. The filter prototype is implemented and measured. The measured results validate the theoretical analysis and show that the filter has a sharp skirt and an out‐of‐band rejection level as good as 38 dB. Meanwhile, return loss is better than 16 dB. © 2015 Wiley Periodicals, Inc. Int J RF and Microwave CAE 25:815–824, 2015.     

基于小波变换的图像条带噪声去除方法

条带噪声的存在不但妨碍高光谱图像的目视判读,而且制约高光谱遥感的定量应用。针对小波变换法条带噪声去除过程中遇到的条带噪声和图像有用信息难以有效分离的问题,根据小波变换的方向性和数学显微镜特性,提出了一种新的基于小波变换的条带噪声去除方法。这种方法首先对含有条带噪声的图像进行一定层数的小波分解;然后对每一层分解得到的与条带噪声分布方向相同的子图像再进行一定层数的小波分解,从而实现条带噪声和图像有用信息的有效分离,将含有条带噪声的子图像置零;最后利用小波反变换得到去除条带噪声的图像。以欧洲空间局PROBA卫星上搭载的CHRIS高光谱数据为例,采用相关系数(R)、结构相似度(SSIM)和峰值信噪比(PSNR)3个定量指标,对比分析了新方法与矩匹配法、傅立叶滤波法和小波阈值法的条带噪声去除效果。结果表明新方法去噪后的图像具有最高的R、SSIM和PSNR,新方法能够有效地去除高光谱图像中的条带噪声,同时较好地保留了原始图像的有用信息。
     

CT图像SART重建技术的CUDA并行实现

在计算机断层扫描(CT)图像重建领域,当投影数据不完备或者含有噪声时,相对于滤波反投影(FBP)算法,联合代数重建方法(SART)能重建出质量更高、更符合临床诊断要求的图像。但SART方法非常耗时,而算法的并行实现是解决这一问题的有效途径之一。提出一种基于nVIDIA通用设备计算架构(CUDA)实现的SART并行运算方法。实验结果表明,该方法在不牺牲重建图像质量的基础上,重建时间大为缩减,更有利于临床应用。     

一种将需求模型转换为软件体系结构模型的方法

需求模型到软件体系结构(SA)模型的转换是软件工程领域的一个研究热点。针对DERM所建立的ACP风格的需求模型,以扩展Petri网风格的SA模型为模型转换的目标,以行为映射为模型转换的依据,提出了一种将需求模型转换为SA模型的方法。首先,讨论了模型转换的整体思路;其次,将行为特征模型中的各个结点转换为SA模型中的构件和连接件,并提出了相应的转换规则;再次,讨论了属性特征模型中主动属性特征的转换,并提出了子系统划分的策略;最后,通过案例研究展示了该方法在从需求模型到SA模型的转换中的有效性。     

程序自动并行化系统

有效的自动并行化系统能帮助用户充分地利用并行计算机资源．本文介绍了自动并行化的历史及现状，并着重介绍了作者所开发的自动并行化系统ＡＦＴ（ａｕｔｏｍａｔｉｃＦｏｒｔｒａｎｔｒａｎｓｆｏｒｍｅｒ）及其特色和所用的关键技术．测试结果表明ＡＦＴ系统在功能上已超越了传统的自动并行化系统．