基于流形正则化的文档分类算法研究

基于流形正则化框架提出一种分类算法(MI_I}RI_SC),以解决高维文档分类问题。该算法通过构建训练样 本的最近部图来佑计数据空间的几何结构并将其作为流形正则化项,结合多变量线性回归获得高维文档的低维流形 结构,并采用k近部分类器对低维流形进行分类,得到针对多类问题的分类器。该算法能够充分利用训练样本的类别 信息来帮助学习以提取有效特征。通过在Rcutcrs 21578数据集上的实验,证明该算法的分类性能和运行速度比传统 分类器有较大的提高。     

基于NPE的Web文本分类方法研究

提出一种基于流形学习的文本分类方法以解决高维文本数据分类问题.利用近邻保持嵌入流形学习算法获得高维Web文本空间中的低维流形结构,采用K近邻分类器对低维流形进行分类.实验结果表明,基于流形学习的方法能获得较好的分类效果,具有稳定的性能.     

基于流形的局部加权协从表示人脸识别

针对稀疏表示分类（Sparse Representation Classification,SRC）运行时间过长和协从表示分类（Collaborative Representation Classification,CRC）仅仅利用人脸数据的全局特征的缺点,本文提出一种基于流形的局部加权协从表示方法（Locality Weighted CRC,LWCRC）,并将其应用在人脸识别中。首先将训练样本中位于高维流形空间中的人脸特征进行局部加权投影到低维空间,随后通过Tikhonov正则化阵用训练样本加权表示测试样本,最后通过最小二乘（Residual Least Square,RLS）分类器进行分类。通过在AR、FERET和Extended-Yale b数据库上实验,本文提出的方法优于SRC、CRC、NRS等方法。     

基于自组织映射的流形学习与可视化

针对自组织映射(SOM)在学习和可视化高维数据内在的低维流形结构时容易产生“拓扑缺陷”的这一问题,提出了一种新的流形学习算法--动态自组织映射(DSOM)。该算法按照数据的邻域结构逐步扩展训练数据集合,对网络进行渐进训练,以避免局部极值,克服“拓扑缺陷”问题;同时,网络规模也随之动态扩展,以降低算法的时间复杂度。实验表明,该算法能更加真实地学习和可视化高维数据内在的低维流形结构;此外,与传统的流形学习算法相比,该算法对邻域大小和噪声也更加鲁棒。所提算法的网络规模和训练数据集合都将按照数据内在的邻域结构进行同步扩展,从而能更加简洁并真实地学习和可视化高维数据内在的低维流形结构。     

结合均值漂移的基于图的半监督图像分类

针对基于图的半监督流形正则化图像分类算法需要大量无标记样本训练分类器,空间和时间复杂度高,甚至不能处理大规模图像,且对背景或目标复杂的图像分类错误率较高的问题,提出了结合均值漂移（mean shift）的基于图的半监督流形正则化图像分类算法。该方法对基于图的半监督流形正则化分类算法的改进主要体现在两方面,首先是通过mean shift算法对图像进行了平滑,以平滑后的图像作为分类对象;其次不是利用所有无标记样本,而是只采用少量无标记样本。实验结果表明:图像的平滑使得目标和背景区域的特征更为一致,从而利用较少的样本就可以提高分类器的正确率;同时大大降低了算法的复杂度,使得基于图的半监督分类算法用于分类大规模图像成为可能。     

基于Nyström 低阶近似的半监督流形排序图像检索

实际图像检索过程中,用户提供的相关反馈有限,但存在大量未标记图像数据. 本文在前期半监督流形图像检索工作的基础上,提出一种基于Nystrm低阶 近似的半监督流形排序图像检索方法.通过采用半监督的流形正则化框架, 将图像数据嵌入到低维流形结构中进行分类排序,以充分利用大量未标记数据, 并兼顾分类误差、数据分布的几何结构以及分类函数的复杂性.针对半监督学习速度缓慢的问题, 基于Nystrm低阶近似对学习过程进行加速.在较大规模的Corel图像数据集上进行了检索实验, 实验结果表明该方法能获得较好的效果.     

基于Nystrm低阶近似的半监督流形排序图像检索

实际图像检索过程中,用户提供的相关反馈有限,但存在大量未标记图像数据.本文在前期半监督流形图像检索工作的基础上,提出一种基于Nystr¨om低阶近似的半监督流形排序图像检索方法.通过采用半监督的流形正则化框架,将图像数据嵌入到低维流形结构中进行分类排序,以充分利用大量未标记数据,并兼顾分类误差、数据分布的几何结构以及分类函数的复杂性.针对半监督学习速度缓慢的问题,基于Nystrm低阶近似对学习过程进行加速.在较大规模的Corel图像数据集上进行了检索实验,实验结果表明该方法能获得较好的效果.     

标架丛上的多流形联络学习算法

传统的流形学习算法通常需要较多的训练样本,将所有样本看作一个流形进行学习,并提取判别特征以进行后续的分类等具体应用。然而在很多实际问题中,并不能获得大量的训练样本,因此存在很多只有一个训练样本的情况。文中提出标架丛上联络学习算法,构建出多流形结构,提取出流形与流形间以及单一流形内的判别信息处理样本少的情形。在处理多流形结构数据集时,利用标架丛上的横空间和纵空间学习模型,将高维空间数据投影到横空间以最大化流形与流形间的间隔,同时又在纵空间中保持同一流形内数据的相关结构。最后通过实例验证了本文算法的有效性。     

改进的多流形LLE学习算法

流形学习已成为机器学习和数据挖掘领域的研究热点。比如,算法LLE（Locally Linear Embedding）作为一种非线性降维算法有很好的泛化性能,被广泛地应用于图像分类和目标识别,但其仅仅假设了数据集处于单流形的情况。MM-LLE（Multiple Manifold Locally Linear Embedding）学习算法作为一种考虑多流形情况的改进算法,依然存在几点不足之处。因此,提出改进的MM-LLE算法,通过任意两类间的局部低维流形组合并构建分类器来提高分类精度;同时改进原算法计算最佳维度的方法。通过与算法ISOMAP、LLE以及MM-LLE比较分类精度,实验结果验证了改进算法的有效性。     

基于局部性正则化推广误差界的特征选择算法

特征选择是当前模式识别领域的研究热点.滤波方法和封装方法是特征选择算法中评价特征子集的两种主要策略,但均不能保证其后所设计的分类器的推广性能.针对以上两种策略的不足,首先引入基于样本流形结构的局部性正则化推广误差界.并在此基础上,以局部性正则化推广误差界为评价函数,以局部性正则化分类方法为目标分类器,提出一种混合滤波-封装型特征选择算法.该算法既保持了较高的计算效率,又保证了目标分类器良好的推广性.实验结果表明,新算法具有比对比算法更优的分类性能.     

基于不确定性度量信息融合的团队一致法研究

给出一种新的信息融合方法--基于不确定性度量的团队一致法.这个方法是基于 一个迭代的团队不确定性度量函数,它能使团队成员达到一致;该方法用于融合从多个图象传 感器得到的对同一个目标的关于目标类别不确定性信息,与其它方法相比该方法有效而简单. 在多图象目标识别中有着广泛的应用.还给出了确定权值系数的一般方法.把这种方法用于一 个现成的实例,得到的结果和多数结合算子的结果一致.     

一种研讨信息组织模型及其在研讨厅中的应用*

在群体研讨过程中会产生大量的研讨信息,群体共识则蕴藏在这些研讨信息之中,因此有必要对研讨信息进行分析处理。为此,提出一种面向群体共识涌现的研讨信息组织模型(DISM)。该模型根据Toulmin逻辑对专家发言进行结构化处理,将专家发言分解为根据、论证、模态限定和主张等几个部分。其中,模态限定是一个可量化的部件,反映了专家对主张的态度;根据和论证则为主张提供支持。该模型是对研讨信息的全面描述,可以有效处理研讨信息并支持共识达成,最后阐述了该模型在综合集成研讨厅中的应用。     

群体研讨环境中面向主张的关联规则分析

群体研讨环境的首要目的是支持合作型的群体研讨活动。在群体研讨活动中,将会产生一些研讨信息,因此有必要分析研讨信息中的关联关系,为达成研讨共识提供参考。阐述了聚类分析和关联规则的基本概念,其中简单描述了Aprioritid算法,提出了改进的K-Means聚类算法,最后通过实例针对研讨主题进行聚类分析,将关联规则应用在聚类分析的结果中,以实现关联关系的挖掘。     

A flexible consensus scheme for multicriteria group decision making under linguistic assessments

The present paper proposes a flexible consensus scheme for group decision making, which allows one to obtain a consistent collective opinion, from information provided by each expert in terms of multigranular fuzzy estimates. It is based on a linguistic hierarchical model with multigranular sets of linguistic terms, and the choice of the most suitable set is a prerogative of each expert. From the human viewpoint, using such model is advantageous, since it permits each expert to utilize linguistic terms that reflect more adequately the level of uncertainty intrinsic to his evaluation. From the operational viewpoint, the advantage of using such model lies in the fact that it allows one to express the linguistic information in a unique domain, without losses of information, during the discussion process.The proposed consensus scheme supposes that the moderator can interfere in the discussion process in different ways. The intervention can be a request to any expert to update his opinion or can be the adjustment of the weight of each expert’s opinion. An optimal adjustment can be achieved through the execution of an optimization procedure that searches for the weights that maximize a corresponding soft consensus index.In order to demonstrate the usefulness of the presented consensus scheme, a technique for multicriteria analysis, based on fuzzy preference relation modeling, is utilized for solving a hypothetical enterprise strategy planning problem, generated with the use of the Balanced Scorecard methodology.     

多执行器-传感器网络协作环境监测和治理

多执行器-传感器网络在环境监测、污染治理方面有着得天独厚的优势. 本文以大气污染为例, 讨论了其在搜寻污染源、释放中和剂方面的应用. 采用更简化的质心维诺分割方法使网络中信息不确定性总和达到局部最优, 同时设计了可与其线性叠加或切换的一致性算法, 以实现各执行器任务负载或载荷比的平均分配. 所引入了一致性算法对质心维诺分割所得到的节点配置产生扰动, 能够进一步降低网络中信息的不确定性. 在仿真实例中, 与单纯使用质心维诺分割配置节点位置的方法进行了比较.     

Consistency and consensus models for group decision-making with uncertain 2-tuple linguistic preference relations

Due to the uncertainty of the decision environment and the lack of knowledge, decision-makers may use uncertain linguistic preference relations to express their preferences over alternatives and criteria. For group decision-making problems with preference relations, it is important to consider the individual consistency and the group consensus before aggregating the preference information. In this paper, consistency and consensus models for group decision-making with uncertain 2-tuple linguistic preference relations (U2TLPRs) are investigated. First of all, a formula which can construct a consistent U2TLPR from the original preference relation is presented. Based on the consistent preference relation, the individual consistency index for a U2TLPR is defined. An iterative algorithm is then developed to improve the individual consistency of a U2TLPR. To help decision-makers reach consensus in group decision-making under uncertain linguistic environment, the individual consensus and group consensus indices for group decision-making with U2TLPRs are defined. Based on the two indices, an algorithm for consensus reaching in group decision-making with U2TLPRs is also developed. Finally, two examples are provided to illustrate the effectiveness of the proposed algorithms.     

An attitude-driven web consensus support system for heterogeneous group decision making

Consensus reaching processes are applied in group decision making problems to reach a mutual agreement among a group of decision makers before making a common decision. Different consensus models have been developed to facilitate consensus reaching processes. However, new trends bring diverse challenges in group decision making, such as the modelling of different types of information and of large groups of decision makers, together with their attitude to achieve agreements. These challenges require the capacity to deal with heterogenous frameworks, and the automation of consensus reaching processes by means of consensus support systems. In this paper, we propose a consensus model in which decision makers can express their opinions by using different types of information, capable of dealing with large groups of decision makers. The model incorporates the management of the group’s attitude towards consensus by means of an extension of OWA aggregation operators aimed to optimize the overall consensus process. Eventually, a novel Web-based consensus support system that automates the proposed consensus model is presented.     

协商研讨环境中的可视化技术研究与实现

利用共识值和关注状态两个协商研讨的重要因素,将隐藏在大量文字中的信息可视化的表现出来,构造了协商研讨环境中基于专家发言的＂研讨树＂,根据研讨信息结构,设计了研讨信息可视化、共识达成趋势可视化和研讨结果可视化的显示模式,实现了基于.net技术的群体协商研讨环境,并通过实验分析了应用效果。     

On consensus reaching process based on social network analysis in uncertain linguistic group decision making: Exploring limited trust propagation and preference modification attitudes

This paper explores a limited trust propagation-based consensus model considering individual attitude for preference modification in a social networked setting with uncertain preference information. To examine the construction of complete linkages, and the status of decision makers in group decision making, it is assumed that the group size and network density will affect the scale of mediators in the propagation process, then a definition of limited trust propagation is proposed and the propagation efficiency can be introduced. On this basis, we obtain missing trust relationships and individual centrality in network. In the process of consensus reaching, both the decision maker’s original preference and recommendation advice are considered for flexibly modeling the preference modification process: the individual attitude toward modification is determined by a newly introduced measure of comprehensive relative out-degree centrality, showing the degree of willingness to adjust assessments. When the willingness is too low to reach the preset consensus level, a multi-objective programming model is designed to improve the consensus as much as possible. Moreover, the proposed feedback mechanism narrows the individual acceptable modification range based on the previous adjustment rule, so as to simulate the personalized and targeted decision behavior. To guarantee obtaining a collective aggregated preference in a logical and precise manner, a two-stage optimization model composing of comprehensive relative in-degree centrality-based information aggregation and best consistency-based uncertainty elimination, is proposed. A numerical example and comparative analyses are performed to show the validity and feasibility of the proposed model.     

犹豫模糊信息相似性测度与群一致性测度及群决策应用

针对目前关于犹豫模糊运算与测度的研究中存在的不足,首先给出犹豫模糊熵函数的定义,并将其作为犹豫模糊信息不确定性测度,进而提出犹豫模糊信息特征向量概念,以信息特征向量为出发点对犹豫模糊距离测度和相似性测度展开研究;为优化群决策过程,提出基于完全优先关系的群一致性测度概念并研究其性质;最后,提出基于相似性测度和群一致性测度的群决策方法并结合算例验证所提出方法的有效性.