首页--工业技术论文--自动化技术、计算机技术论文--计算技术、计算机技术论文--计算机软件论文--程序设计、软件工程论文--程序设计论文

I-nice:一种数据聚类的新方法

摘要第4-5页
Abstract第5-6页
Chapter 1 Introduction第18-30页
    1.1 Motivation第19-20页
    1.2 Research problems第20-23页
    1.3 Research objectives and contributions第23-25页
    1.4 Thesis organization第25-30页
Chapter 2 Background and Preliminaries第30-42页
    2.1 Background第30-38页
        2.1.1 Data clustering第31-32页
        2.1.2 Identifying the number of clusters第32-33页
        2.1.3 Selection of initial cluster centers第33-34页
        2.1.4 Pairwise constraints第34页
        2.1.5 Semi-supervised clustering第34-35页
        2.1.6 Data stream clustering第35-36页
        2.1.7 Concept drift第36-37页
        2.1.8 Survival analysis第37页
        2.1.9 Ensemble clustering第37-38页
    2.2 Preliminaries第38-41页
        2.2.1 Gamma mixture model (GMM)第38-39页
        2.2.2 Expectation-maximization (EM) algorithm第39-40页
        2.2.3 Second-order Akaike information criterion (AICc)第40-41页
        2.2.4 k-means algorithm第41页
    2.3 Summary第41-42页
Chapter 3 I-nice: A New Approach for Clustering第42-82页
    3.1 Introduction第42-46页
    3.2 Related work第46-47页
    3.3 I-nice method and I-nice SO algorithm第47-57页
        3.3.1 Distance distributions of observation points第48页
        3.3.2 Model distance distributions with a GMM第48-49页
        3.3.3 Solve the GMM with EM第49-53页
        3.3.4 Select the best-fitted GMM with AICc第53-54页
        3.3.5 Determine the number of clusters in the data第54页
        3.3.6 Select the initial cluster centers第54-55页
        3.3.7 I-nice SO algorithm第55-57页
    3.4 I-nice MO: I-nice with multiple observation points第57-63页
        3.4.1 Method for combining the results of multiple GMMs第57-61页
        3.4.2 Select the initial cluster centers第61页
        3.4.3 Determine the final set of initial cluster centers第61-62页
        3.4.4 Determine the number of clusters第62-63页
        3.4.5 I-nice MO algorithm第63页
    3.5 Complexity analysis of the I-nice algorithms第63-64页
    3.6 Experiments第64-80页
        3.6.1 Datasets第64-66页
        3.6.2 Experimental settings and evaluation methods第66-69页
        3.6.3 Experimental results and analysis第69-80页
            3.6.3.1 Performance in terms of finding the correct number of clusters第70-73页
            3.6.3.2 Performance in terms of improvement of clustering with I-nice-identified initial cluster centers第73-80页
    3.7 Summary第80-82页
Chapter 4 I-nice based Semi-supervised Clustering from Unlabeled Data第82-112页
    4.1 Introduction第82-84页
    4.2 Related work第84-86页
    4.3 Overview of methodology第86-87页
    4.4 Generate pairwise constraints from unlabeled data第87-93页
        4.4.1 Estimate initial clusters第88页
        4.4.2 Generate dense groups from initial clusters第88-89页
        4.4.3 Select must-link and cannot-link constraints第89-93页
    4.5 Semi-supervised clustering for unlabeled data第93-97页
        4.5.1 Generation of pairwise constraints from unlabeled data第93-94页
        4.5.2 Semi-supervised clustering process第94-96页
        4.5.3 Complexity analysis第96-97页
    4.6 Experiments第97-109页
        4.6.1 Experimental setup第97-100页
            4.6.1.1 Datasets第97-98页
            4.6.1.2 Experimental settings第98页
            4.6.1.3 Evaluation criteria第98-99页
            4.6.1.4 Competing methods第99-100页
        4.6.2 Experimental results and analysis第100-109页
            4.6.2.1 Performance of the I-nice method to estimate the ini-tial clusters第100-101页
            4.6.2.2 Selection of pairwise constraints第101-104页
            4.6.2.3 Performance in terms of improvement of clustering on synthetic datasets第104-105页
            4.6.2.4 Performance in terms of improvement of clustering on real-world datasets第105-109页
    4.7 Summary第109-112页
Chapter 5 I-nice based Concept Drift Detection for Cluster Survival Analysis第112-142页
    5.1 Introduction第113-116页
    5.2 Related work第116-118页
    5.3 Problem formulation第118-121页
    5.4 Concept drift detection for cluster survival analysis第121-128页
        5.4.1 I-nice Stream: A data stream clustering algorithm第121-122页
        5.4.2 Concept drift detection第122-125页
        5.4.3 Cluster survival analysis第125-128页
            5.4.3.1 Categorize the clustering patterns第125-126页
            5.4.3.2 Survival analysis of clustering patterns among windows第126-128页
    5.5 Complexity analysis of the concept drift detectionmethod第128-129页
    5.6 Experiments第129-140页
        5.6.1 Experimental setup第129-131页
            5.6.1.1 Datasets第129-130页
            5.6.1.2 Preprocessing of load profile data第130页
            5.6.1.3 Experimental settings第130-131页
        5.6.2 Experimental results and analysis第131-140页
            5.6.2.1 Performance in terms of improvement of clustering with I-nice Stream第131-132页
            5.6.2.2 Identification of clustering patterns among windows第132-134页
            5.6.2.3 Changing behaviors of power consumption patterns第134-140页
    5.7 Summary第140-142页
Chapter 6 I-nice based Semi-supervised Clustering Ensemble for Load Profile Data Analysis第142-168页
    6.1 Introduction第143-145页
    6.2 Related work第145-146页
    6.3 Semi-supervised clustering ensemble framework第146-154页
        6.3.1 Process load data stream with time horizon第146-147页
        6.3.2 Estimate initial cluster centers with the weighted I-nice method第147-149页
            6.3.2.1 Allocate weighted multiple observation points第148页
            6.3.2.2 Select best-fitted models第148页
            6.3.2.3 Estimate the candidate initial centers第148-149页
            6.3.2.4 Select initial cluster centers第149页
        6.3.3 Semi-supervised clustering第149-152页
        6.3.4 Ensemble of semi-supervised clusterings第152-154页
    6.4 Complexity analysis of semi-supervised ensemble clustering framework第154-156页
    6.5 Experiments第156-167页
        6.5.1 Experimental setup第156-159页
            6.5.1.1 Datasets第156-159页
        6.5.2 Experimental settings and evaluation methods第159页
        6.5.3 Experimental results and analysis第159-167页
            6.5.3.1 Performance in terms of finding the correct number of clusters with theI-nice WMO algorithm第160页
            6.5.3.2 Performance in terms of improvement of clustering with the I-nice WMOalgorithm第160-163页
            6.5.3.3 Performance in terms of improvement of clustering with the ISCE frame-work第163-167页
    6.6 Summary第167-168页
Chapter 7 Conclusions and Future Work第168-172页
Bibliography第172-190页
Acknowledgements第190-192页
Publications第192-193页

论文共193页,点击 下载论文
上一篇:从文本类型理论看“动态对等”与当代圣经汉译
下一篇:工作记忆与情景语境对二语记叙性语篇即时预期推理影响的研究