2003 Fiscal Year Final Research Report Summary
Development of Knowledge Acquisition System that can Adapt to Environment Change
| Project/Area Number |
13558034
|
|---|
| 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 | Osaka University |
|---|
Principal Investigator |
MOTODA Hiroshi Institute of Scientific and Industrial Research, Professor, 産業科学研究所, 教授 (00283804)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
SATOH Ken National Institute of Informatics, Foundations of Informatics Research Division, Professor, 情報学基礎研究系, 教授 (00271635)
YOSHIDA Tetsuya Institute of Scientific and Industrial Research, Research Associate, 産業科学研究所, 助手 (80294164)
WASHIO Takashi Institute of Scientific and Industrial Research, Associate Professor, 産業科学研究所, 助教授 (00192815)
TERABE Masahiro Mitsubishi Research Institute, Inc., Safety Engineering and Technology Department, Researcher, 安全科学研究本部, 研究員
|
|---|
| Project Period (FY) |
2001 – 2003
|
| Keywords | Machine Learning / Feature Selection / Case-base Reasoning / Discretization of Numeric Attribute / Knowledge Acquisition / Environmental Change Detection / Knowledge Maintenance / Minimal Cover |
|---|
| Research Abstract |
In this study an attempt is made to integrate a knowledge acquisition technique which is based on the notion of refinement of existing knowledge introducing the finding from cognitive science and an inductive learning technique which has been developed in the field of machine learning to induce a classifier from accumulated data, to propose a new knowledge acquisition technique to fuse these two different knowledge sources into an operational knowledge, and to verify its effectiveness using real world datasets. More concretely, the following study has been conducted: l) to study a method in which there is no need to know how the knowledge has been acquired and stored in the knowledge base and it is assured that the acquisition of new knowledge does not cause the problem of inconsistency with the existing knowledge, 2) to study a method to conduct continuous knowledge acquisition while automatically identifying which pieces of knowledge have become useless and deleting them still maintaining the overall consistency and the understandability of the constructed knowledge base, 3) to study a method to utilize the accumulated data in such a way that switching between two different knowledge sources (i.e. human exert and accumulated data) can be made at any time of knowledge acquisition without rebuilding the knowledge base from scratch and adapt to environment changes. The developed system has been tested against many datasets of different properties and confirmed to exhibit satisfactory performance. It is now possible to start constructing a knowledge base system acquiring initial pieces of knowledge from human expert and then switching to inductive learning later when abundant data have been accumulated. System developer no more need worry about which pieces of knowledge to delete.
|
