2011 Fiscal Year Final Research Report
Mining Structural Changes
| Project/Area Number |
21300047
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Intelligent informatics
|
|---|
| Research Institution | Hokkaido University |
|---|
Principal Investigator |
HARAGUCHI Makoto 北海道大学, 大学院・情報科学研究科, 教授 (40128450)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
TOMITA Etsuji 電気通信大学, 名誉教授 (40016598)
OKUBO Yoshiaki 北海道大学, 大学院・情報科学研究科, 助教 (40271639)
|
|---|
| Project Period (FY) |
2009 – 2011
|
| Keywords | 変化検出問題 / 情報量差分 / 制約マイニング |
|---|
| Research Abstract |
We have targeted patterns of variables whose correlations get increased after an event, while the correlations are uncorrelated before the event. To take into account positive, negative and even partial correlations among variables, we adopt Kullback-Leibler divergence for two contexts, before and after the event, and take their difference as the measure of changes. To reduce the computational cost for set partitioning, we implemented two strategies, Double Clique Constraints and Jumping Emerging Correlation Change, and showed that the correlation change problem can be solved efficiently even for large data sets.
|
