Workflows for Heliophysics

In this paper we describe how we have introduced workflows into the working practices of a community for whom the concept of workflows is very new, namely the heliophysics community. Heliophysics is a branch of astrophysics which studies the Sun and the interactions between the Sun and the planets, by tracking solar events as they travel throughout the Solar system. Heliophysics produces two major challenges for workflow technology. Firstly it is a systems science where research is currently developed by many different communities who need reliable data models and metadata to be able to work together. Thus it has major challenges in the semantics of workflows. Secondly, the problem of time is critical in heliophysics; the workflows must take account of the propagation of events outwards from the sun. They have to address the four dimensional nature of space and time in terms of the indexing of data. We discuss how we have built an environment for Heliophysics workflows building on and extending the Taverna workflow system and utilising the myExperiment site for sharing workflows. We also describe how we have integrated the workflows into the existing practices of the communities involved in Heliophysics by developing a web portal which can hide the technical details from the users, who can concentrate on the data from their scientific point of view rather than on the methods used to integrate and process the data. This work has been developed in the EU Framework 7 project HELIO, and is being disseminated to the worldwide Heliophysics community, since Heliophysics requires integration of effort on a global scale.     

Gear Geometric Design by B-Spline Curve Fitting and Sweep Surface Modelling

In this paper, B-spline curve fitting and sweep surface generation are used for the geometric design of involute gears. Tooth profiles are described by a B-spline formulation based on interpolating data points with first and second derivative constrains. Tooth surfaces are generated by sweeping the B-spline profiles along specified trajectories. This representation scheme enables tooth shapes to be inter-actively controlled by manipulating control polygons and sweep trajectories. A CAD–CAE integration allows the analysis of contact and structural three-dimensional problems for various geometric configurations. In the paper, the methodology is applied to the geometric design of involute pinions of face gear drives.     

基于深度学习的图片中商品参数识别方法

计算机计算性能的提升使得深度学习成为了可能。作为计算机视觉领域的重要发展方向之一的目标检测也开始结合深度学习方法并广泛应用于各行各业。受限于网络的复杂度和检测算法的设计,目标检测的速度和精度成为一个trade-off。目前电商领域的飞速发展产生了大量包含商品参数的图片,使用传统方法难以有效地提取出图片中的商品参数信息。针对这一问题,本文提出了一种将深度学习检测算法和传统OCR技术相结合的方法,在保证了识别速度的同时大大提升了识别的精度。本文研究的问题包括检测模型、针对特定数据训练、图片预处理以及文字识别等。本文首先比较了现有的目标检测算法,权衡其优缺点,然后使用YOLO模型完成检测任务,并针对YOLO模型中存在的不足进行了一定的改进和优化,得到了一个专用于检测图片中商品参数的目标检测模型,最后使用tesseract完成文字提取任务。在将整个流程结合到一起后,我们的系统不仅有着较好的识别精度,而且是高效和健壮的。本文最后还讨论了优势和不足之处,并指出了未来工作的方向。     

COMS电路的参数提取及模型建立

静态时序分析是验证时序是否收敛的重要手段,但它需要准确的时序模型,尤其是全定制电路不能使用一般晶圆工厂提供的时序工艺库。这里介绍了使用synopsys公司的各种工具进行寄生参数的提取,瞬态分析和时序模型的建立,并以传输管为例,详细描述了建立查找表时序模型的流程。     

Toward Web-Scale Workflows for Film Production

The screen business encompasses all creative and management aspects related to film, television, and new media content, from concept to production and distribution. Companies in this industry face increasing competition due to market globalization. To stay competitive, they're turning to contemporary technology-enabled business improvement methods, such as business process management. Despite its potential benefits, the use of workflow systems for automating film production processes is largely unexplored. The authors' case study highlights some of the key challenges that lie ahead for Web-scale workflows for film production.     

Parameter Estimation for Optimal Object Recognition: Theory and Application

Object recognition systems involve parameters such as thresholds, bounds and weights. These parameters have to be tuned before the system can perform successfully. A common practice is to choose such parameters manually on an ad hoc basis, which is a disadvantage. This paper presents a novel theory of parameter estimation for optimization-based object recognition where the optimal solution is defined as the global minimum of an energy function. The theory is based on supervised learning from examples. Correctness and instability are established as criteria for evaluating the estimated parameters. A correct estimate enables the labeling implied in each exemplary configuration to be encoded in a unique global energy minimum. The instability is the ease with which the minimum is replaced by a non-exemplary configuration after a perturbation. The optimal estimate minimizes the instability. Algorithms are presented for computing correct and minimal-instability estimates. The theory is applied to the parameter estimation for MRF-based recognition and promising results are obtained.     

连续语音识别中声学建模的组合聚类算法研究

基于三音子连续语音识别的一个关键问题是在有限训练数据的条件下对大量声学模型参数的鲁棒性估计。为了解决这个问题,有两个主要的上下文相关的聚类算法被提出,它们是合并(Agglomerative Clustering)聚类(AGG)和决策树(Tree-based)聚类(TB) 。本文分析了这两种算法的优缺点,并分别对其进行了改进,然后提出了最大似然框架下组合聚类算法。大词汇量连续语音识别(LVCSR)的实验结果表明,和单一的决策树聚类算法比较,提出的组合聚类算法对识别率有显著的提高。     

基于几何参数的步态识别研究

针对人体行走时在空间呈现出的不同几何模式,介绍了一种基于几何参数的步态识别方法,利用Gabor滤波器在不同角度上的滤波特性,对步态图像滤波的基础上,通过扫描二值图像提取人体的步态几何特征：两脚间的宽度、两腿与人体垂直中线的夹角以及两腿交叉点的坐标;步态的分类采用改进后的动态时间扭曲（DTW）方法进行动态模板匹配,基于CASIA步态库的测试表明,该方法的分类准确率达90％。实验结果表明,基于几何参数的步态识别,可以有效地区分不同的运动主体。     

打磨机器人误差建模与参数辨识

简要介绍了一种实用的机器人位姿误差建模方法一摄动法,并利用此方法建立了打磨机器人位姿误差模型,对利用三坐标测量机测量的误差数据进行了建模补偿。误差参数辨识中,在传统的九线法的基础上,提出了单点法测量机器人的误差参数,克服了九线法中三测量点位置误差对测量结果的影响,提高了测量结果的可信度。实验结果表明,建模补偿后机器人位姿误差平均值降低到初始值的1／5左右,最大值降低到初始值的1／6左右,验证了所建模型的正确性和参数辨识方法的有效性。     

噪声环境下说话人识别的TEO-CFCC特征参数提取方法

针对广泛应用于说话人识别的MFCC特征参数在低信噪比环境下识别正确率急剧下降的问题,提出了一种 TECLCFCC特征参数提取方法。在CFCC特征参数的基础上,首先通过信号相位匹配的方法消除语音噪声,然后在 CFCC特征参数的求取过程中加入Teager能量算子,将语音的能量作为说话人的特征参数之一,得到TECLCFCC特 征参数。实验表明,提出的"I'E+CFCC特征参数在信噪比为一5d>3的汽车噪声条件下,识别正确率可达到83.2000     

3D模型重建中成型特征识别与参数提取技术

以数字化工艺卡片为研究对象,在已知成型特征类型的前提下,提出一种新的面向三维工序模型重建的成型特征识别与参数提取方法。首先对工序图进行连通域的划分,然后采用最小转角法从连通域中提取成型特征的轮廓线,并与预定义特征进行匹配,对于标准类特征采用编码法进行特征识别,随后进行尺寸标注识别,从中提取出成型特征的特征参数,重建成型特征的三维模型。最后,开发了原型系统,并进行了应用验证。     

A Long Memory Pattern Modelling and Recognition System for Financial Time-Series Forecasting

In this paper, the concept of a long memory system for forecasting is developed. Pattern modelling and recognition systems are introduced as local approximation tools for forecasting. Such systems are used for matching the current state of the time-series with past states to make a forecast. In the past, this system has been successfully used for forecasting the Santa Fe competition data. In this paper, we forecast the financial indices of six different countries, and compare the results with neural networks on five different error measures. The results show that pattern recognition-based approaches in time-series forecasting are highly accurate, and that these are able to match the performance of advanced methods such as neural networks. Received: 2 April 1998?Received in revised form: 1 February 1999?Accepted: 16 February 1999     

跨组织工作流的动态协同技术研究

跨组织的工作流致力于跨越组织边界进行业务重组。针对多个组织间动态协同的复杂性,在组织间协作过程中必须考虑对组织的隐私信息、已建立的工作流和已建立的工作流管理系统的保护,解决成员间进行信息共享时出现的问题,从而真正实现协同商务。为满足这些需求,提出了基于视图的方法,这个方法考虑了工作流和资源的部分可见性,不同程度的可见性可以使企业保持其内部工作流隐私和安全性所需的水平,从而把工作流的可见性降低到合作所需的最低水平。研究的目的是使得组织在增强对外交互能力的同时,保护组织内部的敏感信息。     

跨组织工作流的动态协同技术研究

跨组织的工作流致力于跨越组织边界进行业务重组.针对多个组织间动态协同的复杂性,在组织问协作过程中必须考虑对组织的隐私信息、已建立的工作流和已建立的工作流管理系统的保护,解决成员间进行信息共享时出现的问题,从而真正实现协同商务.为满足这些需求,提出了基于视图的方法,这个方法考虑了工作流和资源的部分可见性,不同程度的可见性可以使企业保持其内部工作流隐私和安全性所需的水平,从而把工作流的可见性降低到合作所需的最低水平.研究的目的是使得组织在增强对外交互能力的同时,保护组织内部的敏感信息.     

Exception Handling for Conflict Resolution in Cross-Organizational Workflows

Workflow management systems (WfMSs) are being increasingly deployed to deliver e-business transactions across organizational boundaries. To ensure a high service quality in such transactions, exception-handling schemes for conflict resolution are needed. The conflicts primarily arise due to failure of a task in workflow execution because of underlying application, or controlling WfMS component failures or insufficient user input. So far, little progress has been reported in addressing conflict resolution in cross-organizational business processes, though its importance has been recognized. In this paper, we identify the exception handling techniques that support conflict resolution in cross-organizational settings. In particular, we propose a novel, bundled exception-handling approach, which supports (1) exception knowledge sharing--sharing exception specifications and handling experiences, (2) coordinated exception handling, and (3) intelligent problem solving--using case based reasoning to reuse exception handing experiences. A prototype of this exception handling mechanism is developed and integrated as a part of the METEOR Workflow Management System. An evaluation of our approach is also presented through some sample workflow applications.     

基于SVM的说话人识别参数选择方法

针对支持向量机(SVM)计算复杂度高的问题,采用归一化和主元分析变换算法对语音数据进行预处理,并把K倍交叉验证与网络搜索法相结合应用到语音识别中。分析结果表明,与遗传算法和粒子群优化算法相比,该方法可以在识别率基本不变的情况下有效提高 SVM的参数寻优效率。     

成本约束的网格工作流时间优化方法

针对成本约束有向无环图DAG(directed acyclic graph)表示的网格工作流完工时间最小化问题,提出两个基于优先级规则的迭代启发算法.算法利用并行活动特征定义正向分层和逆向分层两个概念,将其分别引入最大收益规则MP(maximum profit),得到正分层最大收益规则MPTL(maximum profit with top level) 和逆分层最大收益规则MPBL(maximum profit with bottom level).两规则每次迭代尽量以完工时间的最小增加换取总费用的最大降低,逐步将分层初始解构造为满足成本约束的可行解.模拟结果表明,两规则在获得较少迭代次数和运行时间的同时,能显著改进MP规则的平均性能,且MPBL优于MPTL.     

一种基于压缩感知的说话人识别参数分析

本文为在传统的说话人识别理论研究中“较少的特征参数量不能与较高的识别率共存”的难题找到了一种解决方案。本文基于压缩感知的理论,利用行阶梯观测矩阵进行信号的投影,改变了传统的梅尔频率倒谱系数(Mel-frequency cepstral coefficient, MFCC)参数,从而提出了一种新的识别参数CS-MFCC（Compressed sensing-MFCC）。该参数不仅使得参数存储量降低到少于原存储量的1/n（n为行阶梯观测矩阵的压缩比）,而且明显提高了系统的鲁棒性。通过仿真 实验证明了当压缩比n为4时,平均识别率能够提高到96%以上。     

Sweep methods for parallel computational geometry

In this paper we give efficient parallel algorithms for a number of problems from computational geometry by using versions of parallel plane sweeping. We illustrate our approach with a number of applications, which include:

General hidden-surface elimination (even if the overlap relation contains cycles).

CSG boundary evaluation.

Computing the contour of a collection of rectangles.

Hidden-surface elimination for rectangles.

There are interesting subproblems that we solve as a part of each parallelization. For example, we give an optimal parallel method for building a data structure for line-stabbing queries (which, incidentally, improves the sequential complexity of this problem). Our algorithms are for the CREW PRAM, unless otherwise noted.     

网格环境下工作流的费用-时间调度算法

开放网格服务架构（OGSA）和计算经济模型的提出,使得动态的、不同QoS的服务支持下的资源调度成为一个复杂且具有挑战性的问题。该文提出了网格环境下基于费用-时间的工作流调度算法,该算法采用动态资源选择策略适应网格计算环境下的动态性和自治性。在追求较小的工作流完成时间的同时,对费用进行了优化。模拟结果显示该调度算法符合计算网格的复杂环境,能够更好地满足不同用户的实际需要。