Dissertation > Excellent graduate degree dissertation topics show

Research on Co-clustering and Application

Author: WangYue
Tutor: ZhangZhongFei
School: Zhejiang University
Course: Electronics and Communication Engineering
Keywords: Joint clustering Non - negative matrix factorization EM algorithm Text Mining
CLC: TP311.13
Type: Master's thesis
Year: 2012
Downloads: 82
Quote: 0
Read: Download Dissertation

Abstract


Clustering analysis techniques to the similarity between the basis of the object of study, the object will have a similar pattern in the vast data set gathered into a plurality of different classes. Years, cluster analysis by experts and scholars at home and abroad in-depth study and learning, many excellent, made a lot of good results, so this effect is significantly powerful data mining analysis technology has been a great development . In recent years, with the ever-changing computer technology, the rapid development of the Internet industry, data information has become increasingly diverse, increasingly large scale, people gradually found, based on a single type of clustering technology due to its inherent scalability poor, to handle multiple types of data scarcity shortcomings, has become increasingly unable to meet the needs of users. In this context, the two types and even multi-joint type data clustering technology came into being. Multi-joint clustering technology in recent years has attracted more and more eyeballs, the technology is widely used, can play a great role in gene analysis, search engine, e-commerce and other fields, but its development is still a big limitations and immaturity sex. Research this article on this subject, mainly to do the work in four areas: (1) a brief introduction to the historical background of the cluster analysis, research significance as well as domestic and foreign research status, in-depth analysis of the development of the cluster analysis of the existing technology careful analysis of the advantages and disadvantages of these technologies. (2) Based on the analysis and understanding of the already excellent clustering technology, this paper establishes a model based on the EM iterative update non-negative matrix factorization (Tri-NMF) The model combines a complex spectrum divided principle and based on the strengths of the criteria for division principle, while adding weight adjustment factor, so that the model combines the advantages of both, but also flexible for different data. (3) The theoretical basis of this model, the establishment of a Tri-NMF joint model-based clustering algorithm family include two types and even multi-joint type data clustering the hard analysis method and soft analysis methods. (4) In order to verify the effectiveness and practicality of the system, this paper extracted two standard data sets were fully detailed experiments. The experimental results show that the accuracy (AC) and normalized mutual information (NMI) two classic is widely used as the clustering analysis techniques to measure the performance of the proposed joint clustering method family is better than other several excellent existing clustering techniques. These are proved in this paper based on the joint clustering algorithm Tri-NMF model validity and correctness of the family, and good scalability can, so it has good practical value and application prospect.

Related Dissertations

  1. Research on Sentiment Orientation Analysis of Blog Article Based on Blog Search,TP391.1
  2. The Study of Topic-Oriented IT News with Search Enging and Web Page Analysing,TP393.092
  3. Counting interval mapping of quantitative trait,O212.7
  4. Non-negative Local Coordinate Factorization for Image Representation,TP391.41
  5. Non-negative matrix factorization based on sparse image retrieval,TP391.41
  6. Research on Human Abnormal Behavior Detection and Recognition in Intelligent Video Surveillance,TP391.41
  7. Chinese Movies in the Vision of American Movies Critics,J905
  8. Internet Oriented Sentiment Orientation Analysis of Chinese Public Opinion,TP391.1
  9. Hyperspectral Unmixing Based on Nonnegative Matrix Factorization,TP751.1
  10. Text -oriented disciplines correlation analysis association rule mining technology research,TP311.13
  11. Of speckle noise removal in medical ultrasound image,TP391.41
  12. Distributed sound source localization and tracking algorithm,TN912.3
  13. Investigation on Channel Estimation Based on Pilot Arrangement in MIMO-OFDM System,TN919.3
  14. Research on Hyperspectral Remote Sensing Image Fusion and Quality Evaluation Method,TP751
  15. Sensor Network Interface and Multisensor Fusion Technology; Sensor Network Interfaceand Multisensor Fusion Technology,TP212.9
  16. Network Anomaly Detection and Analysis Base on Sliding Window Wavelet Binary Tree,TP393.08
  17. Spam Filtering System Study Based on Non-negative Matrix Factorization,TP393.098
  18. The application of data mining in the acupuncture literature Meridian Research in,TP311.13
  19. Research and Implementation on Key Technologies of Web Text Mining Oriented to Enterprise Competitive Intelligence,TP391.1
  20. Research on Antomatic Chinese Text Summarization of Web-oriented Text Mining,TP391.1

CLC: > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer software > Program design,software engineering > Programming > Database theory and systems
© 2012 www.DissertationTopic.Net  Mobile