Software execution processes as an evolving complex network

Inspired by the surprising discovery of several recurring structures in various complex networks, in recent years a number of related works treated software systems as a complex network and found that software systems might expose the small-world effects and follow scale-free degree distributions. Different from the research perspectives adopted in these works, the work presented in this paper treats software execution processes as an evolving complex network for the first time. The concept of software mirror graph is introduced as a new model of complex networks to incorporate the dynamic information of software behavior. The experimentation paradigm with statistical repeatability was applied to three distinct subject programs to conduct several software experiments. The corresponding experimental results are analyzed by treating the software execution processes as an evolving directed topological graph as well as an evolving software mirror graph. This results in several new findings. While the software execution processes may demonstrate as a small-world complex network in the topological sense, they no longer expose the small-world effects in the temporal sense. Further, the degree distributions of the software execution processes may follow a power law. However, they may also follow an exponential function or a piecewise power law.     

Shape from Shading Using Probability Functions and Belief Propagation

Shape-from-shading (SFS) aims to reconstruct the three-dimensional shape of an object from a single shaded image. This article proposes an improved framework based on belief propagation for computing SFS. The implementation of the well-known brightness, integrability and smoothness constraints inside this framework is shown. We implement the constraints as probability density functions. For example, the brightness constraint is a two-dimensional probability density function that relates all possible surface gradients at a pixel to their probability given the pixel intensity. A straightforward extension of the framework to photometric stereo is presented, where multiple images of the same scene taken under different lighting conditions are available. The results are promising, especially since the solution is obtained by iteratively applying simple operations on a regular grid of points. The presented framework therefore can be implemented in parallel and is a reasonably likely biological scheme.     

HK网络演化模型的研究和改进

研究一种高集聚度的无标度网络模型,以Holme和Kim（HK）模型为基础,通过引进新增节点所应该具备的连接动态性,改进HK模型的局部特性。分析和仿真表明,改进模型HK’不仅继承了HK模型的高聚类效应和无标度特性,且它的幂律指数可在2～3之间变化,网络成员间的竞争机制也与现实社会网络较接近。     

基于逆Lipschitz条件的二阶导数迭代学习控制

为了提高收敛性和扩展学习律选项,避免高阶相对度的强约束条件,提出了在全局逆Lipschitz条件下的二阶导数迭代学习控制.利用范数理论证明了算法的收敛性.由于二阶导数对噪音敏感,采用多项式滤波避免了振荡,利于工程应用.仿真结果表明,该算法是有效的,应用于不同对象时提高了收敛速度,且精度满足工程应用.     

基于神经网络补偿的转台伺服系统控制研究

针对飞行仿真转台伺服系统中存在的非线性摩擦干扰进行了研究,采用一种基于RBF神经网络进行误差补偿的在线自适应控制策略。在基于逆动力学的计算力矩控制方法的基础上,利用RBF神经网络的万能逼近特性在线辨识模型误差,从而对系统进行补偿,其权值自适应律根据Lyapunov稳定性理论推导,保证了系统跟踪误差的收敛及稳定,仿真结果表明该控制策略可使位置MAE指标从0.0087m提高到0.0016m,使位置MSE指标从1.0128e-4m提高到3.3002e-6m,具有较高的鲁棒性和稳态控制精度。最后分别从隐层节点数及节点中心学习算法的变化两方面提出两种改进方案,仿真结果表明隐层节点数的增加可以进一步减小位置误差,而采用K-means均值聚类算法解决了神经网络节点中心按经验选取或试凑的困难。     

基于半监督典型相关分析的多视图维数约简

为了有效地在半监督多视图情景下进行维数约简,提出了使用非负低秩图进行标签传播的半监督典型相关分析方法。非负低秩图捕获的全局线性近邻可以利用直接邻居和间接可达邻居的信息维持全局簇结构,同时,低秩的性质可以保持图的压缩表示。当无标签样本通过标签传播算法获得估计的标签信息后,在每个视图上构建软标签矩阵和概率类内散度矩阵。然后,通过最大化不同视图同类样本间相关性的同时最小化每个视图低维特征空间类内变化来提升特征鉴别能力。实验表明所提方法比已有相关方法能够取得更好的识别性能且更鲁棒。     

Cross-border issues under EU data protection law with regards to personal data protection

We are living in an inter-connected, global digital society where the services of different operating systems are universal in nature, but many Internet activities are still being tackled by national laws and regulations. A long-existing question is which law is applicable in cases of Internet activities because the online world does not have any physical boundaries. How the European Union (EU) approaches this duality has become a concern for data protection laws. By analysing some recent Court of Justice of the European Union case laws, this article seeks to discover how the EU data protection law tackles disputes involving transnational issues online, which includes its extra-territorial application and cross-border data transfers. The article also indicates that there is an enormous gap between legislation and practice.     

基于mean-shift全局立体匹配方法

针对图像全局立体匹配精度高、计算量大的问题,提出基于mean shift图像分割的全局立体匹配方法。首先,通过mean shift算法对图像进行分割,获取图像同质区域数量和区域的标号。在计算匹配代价时,根据像素所属的分割区域,对像素进行筛选,从而提高匹配代价计算速度;其次,在代价聚合前,将mean shift算法获取的同质区域数K值赋值给K-means聚类算法,对像素再次聚类,提高立体匹配精度和速度;最后通过TRW-S置信传播解决能量最小化问题。实验表明,该算法明显提高了匹配的准确性和速度,与单纯的全局匹配算法相比,具有更大的优势。     

基于PageRank的社交网络影响最大化传播模型与算法研究

社交网络中影响最大化问题是指找出最具有影响力的k个节点,使得最终社交网络中被影响的节点最多,信息传播范围最大。针对影响最大化问题,目前已存在一些基本传播模型,但是这些模型没有考虑网络中节点的相关性和重要性,而网络中节点的相关性和重要性是衡量其影响力的一个重要指标,因此,提出了一种基于网页排名算法的信息传播模型(PageRank-based Propagation Model,PRP),然后利用贪心算法来近似求解影响最大化问题。实验结果表明,基于PageRank的传播模型解决影响最大化问题的效果比传统的线性阈值模型、加权级联模型和独立级联模型的效果更好,影响力范围更大。