改进的加权稀疏表示人脸识别算法

针对传统的加权稀疏表示分类方法在获取训练样本权重以及求解l₁范数最小化问题中计算效率低的问题,提出了一种加权稀疏表示和对偶增广拉格朗日乘子法（DALM）相结合的人脸识别算法WSRC_DALM算法.该算法主要采用高斯核函数计算每个训练样本与测试样本之间的相关性,即获得训练样本相对于测试样本的权重;接着利用DALM算法求解l₁范数最小化模型,实现测试样本的精准重构和分类,最后在ORL和FEI人脸数据集上进行算法验证.在ORL数据集中,WSRC_DALM算法的识别率高达99%,相比经典的SRC和WSRC算法,识别率分别提高了7%和4.8%,同时计算效率比WSRC算法提高了约20倍;在FEI数据集中,多姿态变化下的人脸识别率接近于92%.实验结果表明,WSRC_DALM算法在识别准确度和计算效率上具有明显的优势,并且对较大类内变化具有较好的鲁棒性.     

RB模型实例集上置信传播算法的收敛性

置信传播算法求解RB（k,n,α,r_c,p）模型实例时非常有效,几乎能够有效求解接近可满足性相变点的难解实例.然而,因子图带有回路的实例,置信传播算法不总有效,常表现为不收敛.对于这种现象,至今缺少系统的理论解释.置信传播算法是最为基础的信息传播算法,对置信传播算法的收敛性分析是其他信息传播算法收敛性分析的重要基础.在RB（k,n,α,r_c,p）模型中,取k=2,α>（1/k）,r_c>0均为常数,且满足ke^{-（α/（r_c））}≥1.证明了如果p∈（0,n^-2α）,则置信传播算法在RB（k,n,α,r_c,p）模型产生的随机实例集上高概率收敛.最后,在RB（k,n,α,r_c,p）模型上选取了几组不同的数据进行数值模拟,实验结果表明该结论有效.当问题规模n增大时,在RB（k,n,α,r_c,p）模型的可满足区域,实验收敛区间趋于一个固定范围,而理论收敛区间逐渐变窄.原因在于,RB（k,n,α,r_c,p）模型是一个具有增长定义域的随机CSP实例产生模型,不协调赋值的数目与参数p及问题规模n有关.     

基于不相似性度量优化的密度峰值聚类算法

密度峰值聚类（clustering by fast search and find of density peaks,简称DPC）是一种基于局部密度和相对距离属性快速寻找聚类中心的有效算法.DPC通过决策图寻找密度峰值作为聚类中心,不需要提前指定类簇数,并可以得到任意形状的簇聚类.但局部密度和相对距离的计算都只是简单依赖基于距离度量的相似度矩阵,所以在复杂数据上DPC聚类结果不尽如人意,特别是当数据分布不均匀、数据维度较高时.另外,DPC算法中局部密度的计算没有统一的度量,根据不同的数据集需要选择不同的度量方式.第三,截断距离d_c的度量只考虑数据的全局分布,忽略了数据的局部信息,所以d_c的改变会影响聚类的结果,尤其是在小样本数据集上.针对这些弊端,提出一种基于不相似性度量优化的密度峰值聚类算法（optimized density peaks clustering algorithm based on dissimilarity measure,简称DDPC）,引入基于块的不相似性度量方法计算相似度矩阵,并基于新的相似度矩阵计算样本的K近邻信息,然后基于样本的K近邻信息重新定义局部密度的度量方法.经典数据集的实验结果表明,基于不相似性度量优化的密度峰值聚类算法优于DPC的优化算法FKNN-DPC和DPC-KNN,可以在密度不均匀以及维度较高的数据集上得到满意的结果;同时统一了局部密度的度量方式,避免了传统DPC算法中截断距离d_c对聚类结果的影响.     

不同通信模型下的全光树环网波长分配算法

研究了波分复用全光树环网在不同通信模型下的波长分配算法及其最坏性能分析.对于静态模型,证明了5L/2是树环网所需波长数的紧界.对于动态模型,提出了一种近似比为∑_i=1^hmax_r∈Ri[log|V(r)|]+h的波长分配算法,其中h为树环网的基树的层数,R_i为树环网中处于第i层的环的集合,|V(r)|为环r上的节点数.对于增量模型,提出了一种近似度为O[log₂(t+1)]的波长分配算法,其中t为树环网中的环数.     

有Mate-Pairs的个体单体型MSR问题的参数化算法

个体单体型MSR(minimum SNP removal)问题是指如何利用个体的基因测序片断数据去掉最少的SNP(single-nucleotide polymorphisms)位点,以确定该个体单体型的计算问题.对此问题,Bafna等人提出了时间复杂度为O(2^kn²m)的算法,其中,m为DNA片断总数,n为SNP位点总数,k为片断中洞(片断中的空值位点)的个数.由于一个Mate-Pair片段中洞的个数可以达到100,因此,在片段数据中有Mate-Pair的情况下,Bafna的算法通常是不可行的.根据片段数据的特点提出了一个时间复杂度为O((n-1)(k₁-1)k₂2^2h+(k₁+1)^2h+nk₂+mk₁)的新算法,其中,k₁为一个片断覆盖的最大SNP位点数(不大于n),k₂为覆盖同一SNP位点的片段的最大数(通常不大于19),h为覆盖同一SNP位点且在该位点取空值的片断的最大数(不大于k₂).该算法的时间复杂度与片断中洞的个数的最大值k没有直接的关系,在有Mate-Pair片断数据的情况下仍然能够有效地进行计算,具有良好的可扩展性和较高的实用价值.     

基于Hamming范数的XML流相关性估测算法

在数据库理论中,如何在较小的空间条件下快速地比较不同的XML(extensible markup language)流的差异性是一个基本问题.在这一问题的研究中,人们提出了树编辑距离等测度来描述XML文本的差异性.提出了一种基于Hamming范数的l₀测度——即XML树的不同子树的个数,并以此来刻画XML文本的相关性.在数据流模型下,给出了基于空间有界伪随机数发生器、稳态分布于哈希函数的l₀测度的概率算法.理论上的时空复杂性分析、正确性证明与实验模拟结果表明,这一概率算法对问题的输入提供了一个理想的近似.     

基于增强稀疏性特征选择的网络图像标注

面对网络图像的爆炸性增长,网络图像标注成为近年来一个热点研究内容,稀疏特征选择在提升网络图像标注效率和性能方面发挥着重要的作用.提出了一种增强稀疏性特征选择算法,即,基于l_2,1/2矩阵范数和共享子空间的半监督稀疏特征选择算法(semi-supervised sparse feature selection based on l_2,1/2-matix norm with shared subspace learning,简称SFSLS)进行网络图像标注.在SFSLS算法中,应用l_2,1/2矩阵范数来选取最稀疏和最具判别性的特征,通过共享子空间学习,考虑不同特征之间的关联信息.另外,基于图拉普拉斯的半监督学习,使SFSLS算法同时利用了有标签数据和无标签数据.设计了一种有效的迭代算法来最优化目标函数.SFSLS算法与其他稀疏特征选择算法在两个大规模网络图像数据库上进行了比较,结果表明,SFSLS算法更适合于大规模网络图像的标注.     

矩阵条件数及高斯算法平滑分析的进一步研究

算法的复杂度平滑分析是对许多算法在实际应用中很有效但其最坏情况复杂度却很糟这一矛盾给出的更合理的解释.高性能计算机被广泛用于求解大规模线性系统及大规模矩阵的分解.求解线性系统的最简单且容易实现的算法是高斯消元算法(高斯算法).用高斯算法求解n个方程n个变量的线性系统所需要的算术运算次数为O(n³).如果这些方程中的系数用m位表示,则最坏情况下需要机器位数mn位来运行高斯算法.这是因为在消元过程中可能产生异常大的中间项.但大量的数值实验表明,在实际应用中,需要如此高的精度是罕见的.异常大的矩阵条件数和增长因子是导致矩阵A病态,继而导致解的误差偏大的主要根源.设-A为任意矩阵,A是-A受到微小幅度的高斯随机扰动所得到的随机矩阵,方差σ²≤1.Sankar等人对矩阵A的条件数及增长因子进行平滑分析,证明了Pr[K(A)≥α]≤(3.64n(1+4√log(α)))/ασ.在此基础上证明了运行高斯算法输出具有m位精度的解所需机器位数的平滑复杂度为m+71og₂(n)+3log₂(1/σ)+log₂log₂n+7.在上述结果的证明过程中存在错误,将其纠正后得到以下结果:m+71og₂n+3log₂(1/σ)+4√2+log₂n+log₂(1/σ)+7.367.通过构造两个分别关于矩阵范数和随机变量乘积的不等式,将关于矩阵条件数的平滑分析结果简化到Pr[K(A)≥α]≤(6√2n²)/α·σ.部分地解决了Sankar等人提出的猜想:Pr[K(A)≥α]≤O(n/α·σ).并将运行高斯算法输出具有m位精度的解所需机器位数的平滑复杂度降低到m+81og₂n+3log₂(1/σ)+7.实验结果表明,所得到的平滑复杂度更好.     

具有模态词□φ=□1V□2φ且可靠与完备的公理系统

提出具有模态词□φ=□₁V□₂φ的命题模态逻辑,给出其语言、语法与语义,其公理化系统是可靠与完备的,其中,□₁与□₂是给定的模态词.该逻辑的公理化系统具有与公理系统S5相似的语言,但具有不同的语法与语义.对于任意的公式φ,□φ=□₁V□₂φ;框架定义为三元组W,R₁,R₂,模型定义为四元组W,R₁,R₂,I;在完备性定理证明过程中,需要在由所有极大协调集所构成的集合上构造出两个等价关系,其典型模型的构建方法与经典典型模型的构建方法不同.如果□₁的可达关系R₁等于□₂的可达关系R₂,那么该逻辑的公理化系统变成S5.     

优先级驱动的短任务优先RTOS进程调度算法

针对目前普遍采用的基于静态优先级轮转调度算法的不足,提出一种静态优先级驱动的短任务优先动态时间片轮转调度算法,称之为LC调度算法。该算法采用短任务优先的调度策略,动态分配时间片,为每个静态优先级设置两个就绪队列RQ₀和RQ₁,将burst time短的进程插入RQ₁,长的插入RQ₀。当RQ₀上有进程等待时间过长则会自动提升到相应的RQ₁上。模拟实验表明：LC算法在各种条件下都能保持优秀的性能,它充分考虑了调度的开销、性能、响应速度和公平性,并且对burst time预测精度不敏感,相对于其他改进的RR算法具有更好的表现。     

Robust processing techniques for voice conversion

Differences in speaker characteristics, recording conditions, and signal processing algorithms affect output quality in voice conversion systems. This study focuses on formulating robust techniques for a codebook mapping based voice conversion algorithm. Three different methods are used to improve voice conversion performance: confidence measures, pre-emphasis, and spectral equalization. Analysis is performed for each method and the implementation details are discussed. The first method employs confidence measures in the training stage to eliminate problematic pairs of source and target speech units that might result from possible misalignments, speaking style differences or pronunciation variations. Four confidence measures are developed based on the spectral distance, fundamental frequency (f₀) distance, energy distance, and duration distance between the source and target speech units. The second method focuses on the importance of pre-emphasis in line-spectral frequency (LSF) based vocal tract modeling and transformation. The last method, spectral equalization, is aimed at reducing the differences in the source and target long-term spectra when the source and target recording conditions are significantly different. The voice conversion algorithm that employs the proposed techniques is compared with the baseline voice conversion algorithm with objective tests as well as three subjective listening tests. First, similarity to the target voice is evaluated in a subjective listening test and it is shown that the proposed algorithm improves similarity to the target voice by 23.0%. An ABX test is performed and the proposed algorithm is preferred over the baseline algorithm by 76.4%. In the third test, the two algorithms are compared in terms of the subjective quality of the voice conversion output. The proposed algorithm improves the subjective output quality by 46.8% in terms of mean opinion score (MOS).     

Robust fault diagnosis and fault‐tolerant control for non‐Gaussian uncertain stochastic distribution control systems

The purpose of fault diagnosis of stochastic distribution control systems is to use the measured input and the system output probability density function to obtain the fault estimation information. A fault diagnosis and sliding mode fault‐tolerant control algorithms are proposed for non‐Gaussian uncertain stochastic distribution control systems with probability density function approximation error. The unknown input caused by model uncertainty can be considered as an exogenous disturbance, and the augmented observation error dynamic system is constructed using the thought of unknown input observer. Stability analysis is performed for the observation error dynamic system, and the H_∞ performance is guaranteed. Based on the information of fault estimation and the desired output probability density function, the sliding mode fault‐tolerant controller is designed to make the post‐fault output probability density function still track the desired distribution. This method avoids the difficulties of design of fault diagnosis observer caused by the uncertain input, and fault diagnosis and fault‐tolerant control are integrated. Two different illustrated examples are given to demonstrate the effectiveness of the proposed algorithm. Copyright © 2016 John Wiley & Sons, Ltd.     

Discovering similar Chinese characters in online handwriting with deep convolutional neural networks

A primary reason for performance degradation in unconstrained online handwritten Chinese character recognition is the subtle differences between similar characters. Various methods have been proposed in previous works to address the problem of generating similar characters. These methods are basically comprised of two components—similar character discovery and cascaded classifiers. The goal of similar character discovery is to make similar character pairs/sets cover as many misclassified samples as possible. It is observed that the confidence of convolutional neural network (CNN) is output by an end-to-end manner and it can be understood as one type of probability metric. In this paper, we propose an algorithm by leveraging CNN confidence for discovering similar character pairs/sets. Specifically, a deep CNN is applied to output the top ranked candidates and the corresponding confidence scores, followed by an accumulating and averaging procedure. We experimentally found that the number of similar character pairs for each class is diverse and the confusion degree of similar character pairs is varied. To address these problems, we propose an entropy- based similarity measurement to rank these similar character pairs/sets and reject those with low similarity. The experimental results indicate that by using 30,000 similar character pairs, our method achieves the hit rates of 98.44 and 98.05 % on CASIA-OLHWDB1.0 and CASIA-OLHWDB1.0–1.2 datasets, respectively, which are significantly higher than corresponding results produced by MQDF-based method (95.42 and 94.49 %). Furthermore, recognition of ten randomly selected similar character subsets with a two-stage classification scheme results in a relative error reduction of 30.11 % comparing with traditional single stage scheme, showing the potential usage of the proposed method.     

Fast and scalable vector similarity joins with MapReduce

Vector similarity join, which finds similar pairs of vector objects, is a computationally expensive process. As its number of vectors increases, the time needed for join operation increases proportional to the square of the number of vectors. Various filtering techniques have been proposed to reduce its computational load. On the other hand, MapReduce algorithms have been studied to manage large datasets. The recent improvements, however, still suffer from its computational time and scalability. In this paper, we propose a MapReduce algorithm FACET(FAst and sCalable maprEduce similariTy join) to efficiently solve the vector similarity join problem on large datasets. FACET is an all-pair exact join algorithm, composed of two stages. In the first stage, we use our own novel filtering techniques to eliminate dissimilar pairs to generate non-redundant candidate pairs. The second stage matches candidate pairs with the vector data so that similar pairs are produced as the output. Both stages employ parallelism offered by MapReduce. The algorithm is currently designed for cosine similarity and Self Join case. Extensions to other similarity measures and R-S Join case are also discussed. We provide the I/O analysis of the algorithm. We evaluate the performance of the algorithm on multiple real world datasets. The experiment results show that our algorithm performs, on average, 40 % upto 800 % better than the previous state-of-the-art MapReduce algorithms.     

Model checking for a probabilistic branching time logic with fairness

We consider concurrent probabilistic systems, based on probabilistic automata of Segala & Lynch [55], which allow non-deterministic choice between probability distributions. These systems can be decomposed into a collection of “computation trees” which arise by resolving the non-deterministic, but not probabilistic, choices. The presence of non-determinism means that certain liveness properties cannot be established unless fairness is assumed. We introduce a probabilistic branching time logic PBTL, based on the logic TPCTL of Hansson [30] and the logic PCTL of [55], resp. pCTL [14]. The formulas of the logic express properties such as “every request is eventually granted with probability at least p”. We give three interpretations for PBTL on concurrent probabilistic processes: the first is standard, while in the remaining two interpretations the branching time quantifiers are taken to range over a certain kind of fair computation trees. We then present a model checking algorithm for verifying whether a concurrent probabilistic process satisfies a PBTL formula assuming fairness constraints. We also propose adaptations of existing model checking algorithms for pCTL [4, 14] to obtain procedures for PBTL under fairness constraints. The techniques developed in this paper have applications in automatic verification of randomized distributed systems. Received: June 1997 / Accepted: May 1998     

Model Checking with Strong Fairness

In this paper we present a coherent framework for symbolic model checking of linear-time temporal logic (ltl) properties over finite state reactive systems, taking full fairness constraints into consideration. We use the computational model of a fair discrete system (fds) which takes into account both justice (weak fairness) and compassion (strong fairness). The approach presented here reduces the model-checking problem into the question of whether a given fds is feasible (i.e. has at least one computation). The contribution of the paper is twofold: On the methodological level, it presents a direct self-contained exposition of full ltl symbolic model checking without resorting to reductions to either μ-calculus or ctl. On the technical level, it extends previous methods by dealing with compassion at the algorithmic level instead of either adding it to the specification, or transforming compassion to justice. Finally, we extend ctl^∗ with past operators, and show that the basic symbolic feasibility algorithm presented here, can be used to model check an arbitrary ctl^∗ formula over an fds with full fairness constraints. This research was supported in part by an infra-structure grant from the Israeli Ministry of Science and Art, a grant from the U.S.-Israel Binational Science Foundation, and a gift from Intel.     

可信机器学习的公平性综述

人工智能在与人类生活息息相关的场景中自主决策时,正逐渐面临法律或伦理的问题或风险.可信机器学习是建立安全人工智能系统的核心技术,是人工智能领域的热门研究方向,而公平性是可信机器学习的重要考量.公平性旨在研究机器学习算法决策对个人或群体不存在因其固有或后天属性所引起的偏见或偏爱.从公平表征、公平建模和公平决策这3个角度出...     

Generalized Hough Transform Using Regions with Homogeneous Color

A novel generalized Hough transform algorithm which makes use of the color similarity between homogeneous segments as the voting criterion is proposed in this paper. The input of the algorithm is some regions with homogeneous color. These regions are obtained by first pre-segmenting the image using the morphological watershed algorithm and then refining the resultant outputs by a region merging algorithm. Region pairs belonging to the object are selected to generate entries of the reference table for the Hough transform. Every R-table entry stores a relative color between the selected region pairs. This is done in order to compute the color similarity and in turn generate votes during the voting process and some relevant information to recover the transformation parameters of the object. Based on the experimental results, our algorithm is robust to change of illumination, occlusion and distortion of the segmentation output. It recognizes objects which were translated, rotated, scaled and even located in a complex environment.     

To be fair or efficient or a bit of both

Introducing a new concept of (α,β)-fairness, which allows for a bounded fairness compromise, so that a source is allocated a rate neither less than 0α1, nor more than β1, times its fair share, this paper provides a framework to optimize efficiency (utilization, throughput or revenue) subject to fairness constraints in a general telecommunications network for an arbitrary fairness criterion and cost functions. We formulate a non-linear program (NLP) that finds the optimal bandwidth allocation by maximizing efficiency subject to (α,β)-fairness constraints. This leads to what we call an efficiency–fairness function, which shows the benefit in efficiency as a function of the extent to which fairness is compromised. To solve the NLP we use two algorithms. The first is a well-known branch-and-bound-based algorithm called Lipschitz Global Optimization and the second is a recently developed algorithm called Algorithm for Global Optimization Problems (AGOP).We demonstrate the applicability of the framework to a range of examples from sharing a single link to efficiency fairness issues associated with serving customers in remote communities.     

The <Emphasis Type="Italic">k</Emphasis>-Nearest Neighbour Join: Turbo Charging the KDD Process

The similarity join has become an important database primitive for supporting similarity searches and data mining. A similarity join combines two sets of complex objects such that the result contains all pairs of similar objects. Two types of the similarity join are well-known, the distance range join, in which the user defines a distance threshold for the join, and the closest pair query or k-distance join, which retrieves the k most similar pairs. In this paper, we propose an important, third similarity join operation called the k-nearest neighbour join, which combines each point of one point set with its k nearest neighbours in the other set. We discover that many standard algorithms of Knowledge Discovery in Databases (KDD) such as k-means and k-medoid clustering, nearest neighbour classification, data cleansing, postprocessing of sampling-based data mining, etc. can be implemented on top of the k-nn join operation to achieve performance improvements without affecting the quality of the result of these algorithms. We propose a new algorithm to compute the k-nearest neighbour join using the multipage index (MuX), a specialised index structure for the similarity join. To reduce both CPU and I/O costs, we develop optimal loading and processing strategies.