2009 Fiscal Year Final Research Report
Analysis of Learning Machines with Information Geometry and Information Theory
Project/Area Number |
18300078
|
Research Category |
Grant-in-Aid for Scientific Research (B)
|
Allocation Type | Single-year Grants |
Section | 一般 |
Research Field |
Sensitivity informatics/Soft computing
|
Research Institution | Nara Institute of Science and Technology (2008-2009) Kyoto University (2006-2007) |
Principal Investigator |
IKEDA Kazushi Nara Institute of Science and Technology, 情報科学研究科, 教授 (10262552)
|
Co-Investigator(Kenkyū-buntansha) |
IWATA Kazunori 広島市立大学, 情報科学研究科, 助教 (20405492)
|
Project Period (FY) |
2006 – 2009
|
Keywords | 機械学習 / 情報幾何学 / 情報理論 / 神経情報処理 / パターン認識 |
Research Abstract |
We analyzed some learning machines such as support vector machines from the geometrical viewpoint. As results, we proposed a low-complexity SVM with theoretical background and give an analysis to dual-structured learning machines and self-organizing map. Information theory, on the other hand, revealed convergence properties of reinforcement learning and is applied to pattern recognition problems such as dissimilarity. Machine learning methods are also applied to human modeling such as drivers assist systems.
|