Knowledge learning and understanding from incomplete data based on pattern similarity

2009 Fiscal Year Final Research Report

• Project/Area Number: 19500128
• Research Category: Grant-in-Aid for Scientific Research (C)
• Allocation Type: Single-year Grants
• Section: 一般
• Research Field: Intelligent informatics
• Research Institution: The University of Aizu
• Principal Investigator: ZHAO Qiangfu The University of Aizu, コンピュータ理工学部, 教授 (90260421)
• Co-Investigator(Kenkyū-buntansha): LIU YONG 会津大学, コンピュータ理工学部, 上級準教授 (60325967)
• Project Period (FY): 2007 – 2009
• Keywords: 学習と発見

Research Abstract

To acquire understandable knowledge through machine learning, we have proposed the nearest neighbor classification tree (NNC-Tree) and an induction method. NNC-Tree is a kind of multivariate decision trees based on pattern similarity. In this project, we have proposed (1) a method for selecting important features through learning ; (2) a method for unifying the learning algorithms through data fuzzification ; and (3) a method for efficient dimensionality reduction. With these new contributions, we can induce multivariate decision trees more efficiently.

Research Products

• [Journal Article] Learning to detect spam: Naive-Euclidean approach. (2009)
  • Author(s): T. Chan, J. Ji, Q.F. Zhao
  • Journal Title: International Journal of Signal Processing, Image Processing and Pattern Recognition Vol.1,No.1 Pages: 31-38
• [Journal Article] A fast algorithm for inducing neural network trees. (2008)
  • Author(s): H. Hayashi, Q.F. Zhao
  • Journal Title: Journal of Information Processing Vol.49,No.8 Pages: 2878-2889
• [Journal Article] Clustering large sparse text data: a comparative advantage approach
  • Author(s): Jie Ji, T. Chan, Q.F. Zhao
  • Journal Title: Journal of Information Processing Society of Japan
• [Journal Article] Applying naive Bayes classifier to document clustering.
  • Author(s): Jie Ji, Q.F. Zhao
  • Journal Title: Journal of Advanced Computational Intelligence and Intelligent Informatics.
• [Presentation] Model reduction of neural network trees based on dimensionality reduction (2009)
  • Author(s): H. Hayashi, Q.F. Zhao
  • Organizer: Proc. of International Joint Conference on Neural Networks (IJCNN2009)
  • Place of Presentation: Atlanta
  • Year and Date: 20090000
• [Presentation] Fast document clustering based on weighted comparative advantage. (2009)
  • Author(s): J. Ji, T. Chan, Q.F. Zhao
  • Organizer: Proc. of IEEE International Conference on Systems, Man and Cybernetics (SMC2009)
  • Place of Presentation: Texas
  • Year and Date: 20090000
• [Presentation] Induction of compact neural network trees through centroid based dimensionality reduction. (2009)
  • Author(s): H. Hayashi, Q.F. Zhao
  • Organizer: Proc. of IEEE International Conference on Systems, Man and Cybernetics (SMC2009)
  • Place of Presentation: Texas
  • Year and Date: 20090000
• [Presentation] Comparative advantage approach to classifying sparse text data. (2009)
  • Author(s): J. Ji, T. Chan, Q.F. Zhao
  • Organizer: Proc. of IEEE International Conference on Computer and Information Technology (CIT2009)
  • Place of Presentation: Xiamen
  • Year and Date: 20090000
• [Presentation] A study on criteria for extracting key terms in document clustering. (2008)
  • Author(s): J. Ji, R. Shindo, Q.F. Zhao, Y. Kunishi
  • Organizer: Proc. IEEE International Conference on Systems, Man and Cybernetics (SMC2008)
  • Place of Presentation: Singapore
  • Year and Date: 20080000
• [Presentation] A Customer Intention Aware System for Document Analysis.
  • Author(s): Jie Ji, Daichi Kunita, Qiangfu Zhao
  • Organizer: Proceeding of IJCNN2010
  • Place of Presentation: Spain
• [Remarks] Y. Watanabe,“Extracting understandable knowledge based on data fuzzification,"Master Thesis of The University of Aizu, Mar. 2009 (supervised by Q. F. Zhao).