A Novel Learning Method Based on Combinatorial Feature of Data

• Project/Area Number: 20800045
• Research Category: Grant-in-Aid for Young Scientists (Start-up)
• Allocation Type: Single-year Grants
• Research Field: Fundamental theory of informatics
• Research Institution: Ishinomaki Senshu University
• Principal Investigator: HARAGUCHI Kazuya Ishinomaki Senshu University, 理工学部, 助教 (80453356)
• Research Collaborator: NAGAMOCHI Hiroshi 京都大学, 大学院・情報学研究科, 教授 (70202231)
• Project Period (FY): 2008 – 2009
• Project Status: Completed (Fiscal Year 2009)
• Budget Amount: ¥3,042,000 (Direct Cost: ¥2,340,000、Indirect Cost: ¥702,000); Fiscal Year 2009: ¥1,482,000 (Direct Cost: ¥1,140,000、Indirect Cost: ¥342,000); Fiscal Year 2008: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000)
• Keywords: アルゴリズム / 機械学習 / 情報可視化 / 人工知能

Research Abstract

In this project, we have aimed at establishing a novel learning method based on combinatorial feature of data. For classification, an essential learning problem, we proposed a learning algorithm based on bipartite graph structure. In computational experiments, we observed that its learning ability is competitive with previous methods and is even superior in some special cases.

Research Products

• [Journal Article] Classification via Visualization of Sample-feature Bipartite Graphs (2011)
  • Author(s): Kazuya Haraguchi, Seok-Hee Hong, Hiroshi Nagamochi
  • Journal Title: Department of Applied Mathematics and Physics, Kyoto University, Technical Reports
• [Journal Article] Visual Analysis of Hierarchical Data Using 2. 5D Drawing with Minimum Occlusion (2010)
  • Author(s): Kazuya Haraguchi, Seok-Hee Hong, Hiroshi Nagamochi
  • Journal Title: Department of Applied Mathematics and Physics, Kyoto University, Technical Reports
• [Journal Article] Multiclass Visual Classifier Based on Bipartite Graph Representation of Decision Tables (2010)
  • Author(s): Kazuya Haraguchi, Seok-Hee Hong, Hiroshi Nagamochi
  • Journal Title: Proc. LION 4 (LNCS 6073) Pages: 169-183
  • NAID: 120005347334
  • Peer Reviewed
• [Journal Article] Multiclass Visual Classifier Based on Bipartite Graph Representation of Decision Tables (2010)
  • Author(s): K.Haraguchi, S.H.Hong, H.Nagamochi
  • Journal Title: Proceedings of LION 4 LNCS 6073 Pages: 169-183
  • NAID: 120005347334
  • Peer Reviewed
• [Journal Article] Bipartite graph representation of multiple decision table classifiers (2009)
  • Author(s): Kazuya Haraguchi, Seok-Hee Hong, Hiroshi Nagamochi
  • Journal Title: Proc. SAGA 2009 (LNCS 5792) Pages: 46-60
  • Peer Reviewed
• [Journal Article] Visualization can improve multiple decision table classifiers (2009)
  • Author(s): Kazuya Haraguchi, Seok-Hee Hong, Hiroshi Nagamochi
  • Journal Title: Proc. MDAI 2009 (ISBN: 978-84-00-08851-4) Pages: 41-52
  • Peer Reviewed
• [Journal Article] Visualization can improve multiple decision table classifiers (2009)
  • Author(s): K.Haraguchi, S.H.Hong, H.Nagamochi
  • Journal Title: Proceedings of MDAI 2009 ISBN 9788400088514(CD-ROM) Pages: 41-52
  • Peer Reviewed
• [Journal Article] Bipartite graph representation of multiple decision table classifiers (2009)
  • Author(s): K.Haraguchi, S.H.Hong, H.Nagamochi
  • Journal Title: Proceedings of SAGA 2009 LNCS 5792 Pages: 46-60
  • Peer Reviewed
• [Presentation] Learning Classifier by Edge Crossing Minimization (2010)
  • Author(s): K.Haraguchi
  • Organizer: International workshop on Multi-dimensional Visualization
  • Place of Presentation: 京都大学
  • Year and Date: 2010-02-23
• [Presentation] Learning Classifier by Edge Crossing Minimization (2010)
  • Author(s): Kazuya Haraguchi, Seok-Hee Hong, Hiroshi Nagamochi
  • Organizer: Int'l workshop on Multi-dimensional Visualization
• [Presentation] Visualized Multiple Decision Table Classifiers without Discretization (2009)
  • Author(s): Kazuya Haraguchi, Seok-Hee Hong, Hiroshi Nagamochi
  • Organizer: 4th Korea-Japan Workshop on Operations Research in Service Science
• [Presentation] Classification by Ordering Data Samples (2008)
  • Author(s): Kazuya Haraguchi, Seok-Hee Hone, Hiroshi Nagamochi
  • Organizer: Kyoto RIMS Workshop on Acceleration and Visualization of Computation for Enumeration Problems
  • Place of Presentation: 京都大学数理解析研究所
  • Year and Date: 2008-09-29
• [Presentation] Classification by Ordering Data Samples (2008)
  • Author(s): Kazuya Haraguchi, Seok-Hee Hong, Hiroshi Nagamochi
  • Organizer: Kyoto RIMS Workshop on Acceleration and Visualization of Computation for Enumeration Problems (AVCEP08)