Foundations of knowledge discovery based on embedding and dimension reduction in high-dimensional feature space

2015 Fiscal Year Final Research Report

• Project/Area Number: 24300060
• Research Category: Grant-in-Aid for Scientific Research (B)
• Allocation Type: Partial Multi-year Fund
• Section: 一般
• Research Field: Intelligent informatics
• Research Institution: Kyushu Institute of Technology
• Principal Investigator: Hirata Kouchi 九州工業大学, 大学院情報工学研究院, 教授 (20274558)
• Co-Investigator(Kenkyū-buntansha): SHINOHARA Takeshi 九州工業大学, 大学院情報工学研究院, 教授 (60154225) ; KUBOYAMA Tetsuji 学習院大学, 計算機センター, 教授 (80302660)
• Project Period (FY): 2012-04-01 – 2016-03-31
• Keywords: 離散構造距離 / 埋め込み / 次元縮小 / 木編集距離 / Taiマッピング / 特徴選択 / ヒルベルト整列

Outline of Final Research Achievements

As the foundation of knowledge discovery based on embedding and dimension reduction in high-dimensional feature space, this research characterizes mathematically a Tai mapping hierarchy consisting of mappings that provide the variations of a tree edit distance, analyzes the time complexity of computing their variations and provides the several results concerned with the hierarchy. Also this research designs and implements the fastest feature selection algorithms Super-CWC and Super-LCC based on consistency in categorical data. Furthermore, this research proposes the similarity search method in high-dimensional feature space based on Hilbert sorting.

Free Research Field

知能情報学