基于自主认知深度时间聚类表示的隔离开关故障诊断方法

为准确识别隔离开关发生的故障,并确定故障类型,保证电网的稳定运行,提出一种基于自主认知的深度时序聚类表示模型(Autonomous-cognition deep temporal clustering representation model,AC-DTCR)对隔离开关的故障进行诊断。在数据量少且类别标签信息不可用的情况下,时间序列聚类是非常好的无监督学习技术,而AC-DTCR模型集成了时间重建和K-means目标,为提高编码器的能力,提出一种假样本生成策略和辅助分类任务,改进集群结构,获得特定于集群的时间表示。根据高压隔离开关故障模拟试验得到的电机电流数据,使用AC-DTCR模型分成四个部分对试验数据进行训练。结果表明,该模型具有良好的分类性能,与传统的分类模型和时间序列聚类模型相比,有更高的准确率,可应用于电力设备故障诊断领域中。

单可表示类的特征

择优模型是目前常识推理领域中最常用的语义结构之一 .在择优模型中有一类重要的模型——单射择优模型 ,近年来 ,众多的学者对该模型类从各种角度进行了研究 ,但一个重要而基本的问题 ,即何种类型的择优后承具有单射择优模型 ,虽经多位学者的研究到现在仍然悬而未解 .目前 ,最好的结果是 Freund在有限语言限制下证明了“择优后承具有单射择优模型当且仅当它满足弱析取合理性”.文中提出择优模型的一种转换 ,将此转换应用于 Kraus,L ehm an及 Magidor提出的 KL M模型上 ,在一般语言框架下证明了“满足弱析取合理性的择优后承必具有单射择优模型”。

类规则表示及其不精确推理算法

知识表示是目前人工智能中一个研究热门,因为现有的各种知识表示方式,虽然都各有其优点,但也各有其弱点,离人们要求的目标距离还远。本文介绍一种新的知识表示,我们称之为类规则表示。所以称之为类规则,一是因为它与规则表示很类似。

基于k-最小表示误差类的表示分类方法

基于表示的分类(representation-based classification,RC)通常使用所有类的训练样本来表示测试样本。然而,是否需要使用全部类来表示测试样本仍有待研究。为此,提出一种两阶段表示分类框架。首先使用RC算法计算测试样本相对于全部类的训练样本的表示系数,找出前k(k≥1)个具有最小表示误差的类;然后利用该k个类的训练样本,再次应用RC算法对测试样本进行表示,并通过从这k个类中找出最小表示误差类来确定测试样本的类别。此外,提出了一种非负加权协同表示分类算法。所提分类框架中的前后两个RC算法可以相同也可以不同。取前后两个RC相同,对五种RC,在五个数据库上进行实验,实验结果表明,所提两阶段表示分类框架大多数情况下能显著提升原RC算法的分类精度。

基于度量的小样本分类方法研究综述

小样本学习旨在让机器像人类一样通过对少量样本的学习达到对事物认知和概括的能力.基于度量的小样本学习方法希望学习一个低维嵌入空间,直接对比查询集合和支持类之间的相似性,分类测试样本.文中针对基于度量的小样本学习方法,尝试从这类方法需要解决的关键问题、类表示学习和相似性度量入手,梳理相关文献.与已有相关综述不同,文中只针对基于度量的小样本学习方法进行更详尽全面的分类,而且从关键问题角度进行分类.最后总结目前代表性工作在常用的图像分类任务数据集上的实验结果,分析现有方法存在的问题,并展望未来工作.

块核范数的RPCA分解与熵权类稀疏的壁画修复

针对图像修复过程中,颜色纹理光学属性分离不彻底,以及在稀疏表示图像修复时字典设计单一,导致壁画图像修复结果易出现结构不连贯和模糊效应等问题,提出了一种基于块核范数的鲁棒主成分分析(robust principal component analysis,RPCA)分解与熵权类稀疏的壁画修复方法。首先,采用提出的基于块核范数的RPCA图像分解算法,将壁画图像分解为结构层和纹理层,利用块核范数进行纹理矫正操作,克服了RPCA结构纹理分离不完全的问题。然后,提出熵加权k-means方法对结构层图像进行聚类,构建得到稀疏子类字典,并通过奇异值分解和分裂Bregman迭代优化的类稀疏修复方法,完成结构层图像的重构。最后,利用双三次插值算法实现对纹理层图像的修复,将修复后的结构层和纹理层进行融合,完成破损壁画的修复。通过对真实敦煌壁画数字化修复,实验结果表明,该算法能够有效地保护壁画图像的边缘和纹理等重要特征信息,无论从视觉效果还是从峰值信噪比等定量评价方面,提出的方法修复效果均优于比较算法,且修复执行效率更高。

概念驱动的小样本判别特征学习方法

小样本学习旨在让模型能够在仅有少量标记数据的新类中进行分类。基于度量学习的方法是小样本学习的一种有效方法,该类方法利用有标签的支持集样本构建类表示,再基于查询样本和类表示的相似性进行分类。因此,如何构建判别性更强的类表示是这类方法的关键所在。多数工作在构建类表示时,忽略了类概念相关信息的挖掘,这样容易引入样本中类别无关信息,从而降低类表示的判别性。为此本文提出一种概念驱动的小样本判别特征学习方法。该方法首先利用类别的语义信息来指导模型挖掘样本中类概念相关信息,进而构建更具判别性的类表示。其次,设计了随机掩码混合机制增加样本的多样性和识别难度,进一步提升类表示的质量。最后对处于决策边界附近的查询样本赋予更大的权重,引导模型关注难样本,从而更好地进行类表示学习。大量实验的结果表明本文提出的方法能够有效提升小样本分类任务的准确率,并且在多个数据集上优于当前先进的算法。

拟遗传代数的一种判定方法

利用[1]中引进的图表示的两个子范畴,给出拟遗传代数的一种判定方法。

现代汉语主谓谓语句的HNC处理研究

主谓谓语句是汉语特有的语言现象,主谓结构做谓语的格式也是汉语最常见、最重要的句式,也是汉语区别于印欧语的重要句式之一。现代汉语语法对这种特殊句式进行了深入的研究,在语法、语义等方面取得了长足进展。本文尝试在HNC理论的总体框架下,对这一特殊句式展开全面、系统的解释,并从计算语言学的角度形成特有的处理规则,便于计算机在原始语料中发现这类语句,最终形成针对主谓谓语句的专家处理系统。

基于三视图的分布式数控系统模型及其实现

将三视图的思想用于分布式数字控制（ＤＮＣ）系统的分析和设计过程中，以面向对象方法作为主视图，建立了一种基于三视图的ＤＮＣ系统形式模型，并对基于这种模型的ＤＮＣ系统体系结构进行了描述；以实体—关系图和事件—响应模型作为辅助视图构建了类表示模型，提出了一种资源可重用的ＤＮＣ系统解决方案。

构建资源可复用的DNC系统

对基于现场总线技术的DNC系统进行了研究并分析其不足之处 ,提出DNC系统的构建 ,以采用面向对象方法为主 ,传统分析方法为辅 ,最大限度地利用原有资源。文中描述了基于现场总线的DNC系统的总体结构 ,通过对DNC系统的形式化描述 ,建立DNC系统的类表示模型 。

Relationship Between Differential Gene Expression and Heterosis During Ear Development in Maize (Zea mays L.)

Maize （Zea raays L.） is one of the most important crops because of the remarkable properties of its hybrid, which is responsible for the high commercial value of hybrid maize. The genetic basis of heterosis （hybrid vigor） is not well understood. A differential display technique was performed to identify genes with differential expression across twelve maize inbred lines and thirty-three hybrids during ear development. An incomplete diallel design was used to investigate the relationship between the global framework of differential gene expression and heterosis. It was found that the genes belonging to MONO pattern （i.e., genes expressed in both parental lines and in hybrid） was the highest in percentage among the total five patterns and illustrated that the properties of differentially expressed genes are not entirely responsible for heterosis. Furthermore,a larger number of differentially expressed genes in hybrid, which serves as a major reservoir for generating novel phenotypes that exhibit heterosis of certain agronomic traits during early development and differentiation of maize ear. Moreover, there were some silent genesin hybrids that are responsible for the arrest or abortion of spikelets and for the increase in kernels weight.

A new discriminative sparse parameter classifier with iterative removal for face recognition

Face recognition has been widely used and developed rapidly in recent years.The methods based on sparse representation have made great breakthroughs,and collaborative representation-based classification(CRC)is the typical representative.However,CRC cannot distinguish similar samples well,leading to a wrong classification easily.As an improved method based on CRC,the two-phase test sample sparse representation(TPTSSR)removes the samples that make little contribution to the representation of the testing sample.Nevertheless,only one removal is not sufficient,since some useless samples may still be retained,along with some useful samples maybe being removed randomly.In this work,a novel classifier,called discriminative sparse parameter(DSP)classifier with iterative removal,is proposed for face recognition.The proposed DSP classifier utilizes sparse parameter to measure the representation ability of training samples straight-forward.Moreover,to avoid some useful samples being removed randomly with only one removal,DSP classifier removes most uncorrelated samples gradually with iterations.Extensive experiments on different typical poses,expressions and noisy face datasets are conducted to assess the performance of the proposed DSP classifier.The experimental results demonstrate that DSP classifier achieves a better recognition rate than the well-known SRC,CRC,RRC,RCR,SRMVS,RFSR and TPTSSR classifiers for face recognition in various situations.

Underdetermined Blind Source Separation of Adjacent Satellite Interference Based on Sparseness

The problem of underdetermined blind source separation of adjacent satellite interference is proposed in this paper. Density Clustering algorithm(DC-algorithm) presented in this article is different from traditional methods. Sparseness representation has been applied in underdetermined blind signal source separation. However, some difficulties have not been considered, such as the number of sources is unknown or the mixed matrix is ill-conditioned. In order to find out the number of the mixed signals, Short Time Fourier Transform(STFT) is employed to segment received mixtures. Then, we formulate the blind source signal as cluster problem. Furthermore, we construct Cost Function Pair and Decision Coordinate System by using density clustering. At the end of this paper, we discuss the performance of the proposed method and verify the novel method based on several simulations. We verify the proposed method on numerical experiments with real signal transmission, which demonstrates the validity of the proposed method.

An expression atlas of miRNAs in Arabidopsis thaliana

MicroRNAs(miRNAs) are small non-coding RNAs that regulate a variety of biological processes. miRNA expression often exhibits spatial and temporal specificity. However, genome-wide miRNA expression patterns in different organs during development of Arabidopsis thaliana have not yet been systemically investigated. In this study, we sequenced small RNA libraries generated from 27 different organ/tissue types, which cover the entire life cycle of Arabidopsis. Analysis of the sequencing data revealed that most miRNAs are ubiquitously expressed, whereas a small set of miRNAs display highly specific expression patterns. In addition, different miRNA members within the same family have distinct spatial and temporal expression patterns. Moreover, we found that some miRNAs are produced from different arms of their hairpin precursors at different developmental stages. This work provides new insights into the regulation of miRNA biogenesis and a rich resource for future investigation of miRNA functions in Arabidopsis.

Representation type of 0-Hecke algebras

In the present paper we determine the representation type of the 0-Hecke algebra of a finite Coxeter group.

Learning embeddings of a heterogeneous behavior network for potential behavior prediction

Potential behavior prediction involves understanding the latent human behavior of specific groups,and can assist organizations in making strategic decisions.Progress in information technology has made it possible to acquire more and more data about human behavior.In this paper,we examine behavior data obtained in realworld scenarios as an information network composed of two types of objects(humans and actions)associated with various attributes and three types of relationships(human-human,human-action,and action-action),which we call the heterogeneous behavior network(HBN).To exploit the abundance and heterogeneity of the HBN,we propose a novel network embedding method,human-action-attribute-aware heterogeneous network embedding(a4 HNE),which jointly considers structural proximity,attribute resemblance,and heterogeneity fusion.Experiments on two real-world datasets show that this approach outperforms other similar methods on various heterogeneous information network mining tasks for potential behavior prediction.

Stable homotopy classification of A_n^4-polyhedra with 2-torsion free homology

We study the stable homotopy types of F_n^4(2)-polyhedra, i.e.,(n- 1)-connected, at most(n+ 4)-dimensional polyhedra with 2-torsion free homologies. We are able to classify the indecomposable F_n^4(2)-polyhedra. The proof relies on the matrix problem technique which was developed in the classification of representations of algebras and applied to homotopy theory by Baues and Drozd(1999, 2001, 2004).

Order and Type of the Generalized Dirichlet Series

For an entire function represented by a generalized dirichlet series, we define its maximal term, maximal modulus, order and type. We use the classical methods to study the relation between order, type and coeFFIcients, exponents, which improve and generalize some results of the dirichlet series with real exponents.