2014 Fiscal Year Final Research Report
Cluster Analysis Using Dissimilarity Based on Attribute Reduction of Rough Set Theory
| Project/Area Number |
23700265
|
|---|
| Research Category |
Grant-in-Aid for Young Scientists (B)
|
| Allocation Type | Multi-year Fund |
|---|
| Research Field |
Sensitivity informatics/Soft computing
|
|---|
| Research Institution | Osaka University |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2011-04-28 – 2015-03-31
|
| Keywords | データマイニング / 機械学習 / カーネル法 / 論理関数 / ラフ集合 |
|---|
| Outline of Final Research Achievements |
Attribute reduction of rough set theory is a methodology to remove irrelevant attributes from a data set, which is based on discernibility/indiscernibility of object sets. In this research, we have proposed two kinds of similarity/dissimilarity of objects based on the discernibility for nominal data sets. Moreover, we have proposed data analysis methods using them. One is dissimilarities for clusters, which are defined by the number of attribute subsets discerning two clusters. The other is kernel functions reflecting discernibility, whose feature spaces are discerning attribute subsets. We have applied those similarities and dissimilarities to clustering and decision rule induction tasks. It is shown by numerical experiments that we can obtain clusters and decision rules balancing classification accuracy and simplicity using proposed approaches.
|
| Free Research Field |
データマイニング
|
