1999 Fiscal Year Final Research Report Summary
Development of Data Mining System Using Binary Decision Diagrams for Knowledge Representation
| Project/Area Number |
09558032
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 展開研究 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | The University of Tokyo |
|---|
Principal Investigator |
MIYANO Satoru The Institute of Medical Science, The University of Tokyo, Professor, 医科学研究所, 教授 (50128104)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
SHINOHARA Ayumi Department of Informatics, Kyushu University, Associate Professor, システム情報科学研究科, 助教授 (00226151)
MARUYAMA Osamu The Institute of Medical Science, The University of Tokyo, Assistant, 医科学研究所, 助手 (20282519)
AKUTSU Tatsuya The Institute of Medical Science, The University of Tokyo, Associate Professor, 医科学研究所, 助教授 (90261859)
SHIMOZONO Shinichi Department of Artificial Intelligence, Kyushu Institute of Technology, 情報工学部, 助教授 (70243988)
SHOUDAI Takayoshi Department of Informatics, Kyushu University, Associate Professor, システム情報科学研究科, 助教授 (50226304)
|
|---|
| Project Period (FY) |
1997 – 1999
|
| Keywords | binary decision diagram / learning / knowledge discovery / pattern matching algorithm / data mining / computational complexity / entropy / genome |
|---|
| Research Abstract |
Discovery is one of the most basic activities in sciences and it has been recognized that knowledge discovery is very important in human intellectual activity. However, this process of knowledge discovery is being replaced or assisted by systems called data minining, systems.
In order to cope with this situation, this research ,vas started to develop a knowledge discovery system based on binary decision diagrams for knowledge representation. Our mathematical analyses proved that it is computationally hard to deal with binary decision diagrams directly for knowledge representation This insight directed us to develop a system which first deals with decision trees and then convert the decision trees to binary decision diagrams. As a result, we have developed a system called Hypothesis Creator which has (1) a parallelized hypothesis generator with the ability of dealing with several ten million attributes and (2) view design system for creating attributes We have also developed a system called Decision Diagram Editor for converting and crerating decision diagrams which encourages human intervention for hypothesis creation.
|
