2012 Fiscal Year Final Research Report
Creative extensions of data mining theory by means of quantum-mechanical labeling
| Project/Area Number |
22300078
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Sensitivity informatics/Soft computing
|
|---|
| Research Institution | Tohoku University |
|---|
Principal Investigator |
TANAKA Kazuyuki 東北大学, 大学院・情報科学研究科, 教授 (80217017)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
INOUE Jun-ichi 北海道大学, 大学院・情報科学研究科, 准教授 (30311658)
|
|---|
| Co-Investigator(Renkei-kenkyūsha) |
TSUDA Koji 産業技術総合研究所, 生命情報工学研究センター, 主任研究員 (90357517)
|
|---|
| Project Period (FY) |
2010 – 2012
|
| Keywords | 数理工学 / 機械学習 / 統計数学 / 情報統計力学 / 確率的情報処理 |
|---|
| Research Abstract |
General frameworks of statistical machine learning systems based on quantum-mechanical extended Markov random fields has been provided by using belief propagation methods. Some novel statistical model selection schemes by using quantum belief propagation methods have been proposed. As one of unexpected research results in our present project proposal, we have established new statistical learning frameworks for some statistical models with the first-order phase transitions in the stand point of constrained maximization ofentropies instead of conventional maximum likelihood estimations.
|
