1995 Fiscal Year Final Research Report Summary
Automated Knowledge Acguisition based on Rough Sets and Resampling Methods
Project/Area Number |
06680343
|
Research Category |
Grant-in-Aid for General Scientific Research (C)
|
Allocation Type | Single-year Grants |
Research Field |
Intelligent informatics
|
Research Institution | Tokyo Medical and Dental University |
Principal Investigator |
TANAKA Hiroshi Tokyo Medical and Dental University Professor, 難治疾患研究所, 教授 (60155158)
|
Co-Investigator(Kenkyū-buntansha) |
TSUMOTO Shusaku Tokyo Medical and Dental University Assistant Professor, 難治疾患研究所, 助手 (10251555)
|
Project Period (FY) |
1994 – 1995
|
Keywords | rough sets / resampling method / knowledge acguisition / machine learning / expert system / rule induction |
Research Abstract |
The one of the most important problems with clinical application of medical expert systems is that the applied knowledge and reasoning are too superficial. Although there have been studeid many researches on "deeper knowledge representation", it is still difflicult to represent medical knowledge fully. In order to solve this problem, machine learning approaches, which extracts medical knowledge from clinical databases, are studied. However, they are difficult to extract probabilistic knolwedge. In this project, the following two kinds of system are developed. The one, which is studied in 1995, studies an extention of original rough set model which extracts knoledge from clinical databaeses. The system is called PRIMEROSE (Probabilistic Rule Induction Method based on Rough Set theory). Then the next project, which is studied in 1996, studies induction of three kins of rules in the diagnoing model of a expert system RHINOS (Rule based Headache and Facial Pain Information Organizing System). The system is called PRIMEROS-REX (Probablistic Rule Induction Method based on Rough Sets and Resampling Methods for Expert System). Both of the system are applied to clinical databases, whose results shows that the induced results well match with rules acquired from medical experts.
|