A study on distance metric learning of data distribution space based on information theoretic criterion
| Project/Area Number |
22800067
|
|---|
| Research Category |
Grant-in-Aid for Research Activity Start-up
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
Statistical science
|
|---|
| Research Institution | Waseda University |
|---|
Principal Investigator |
HINO Hideitsu 早稲田大学, 理工学術院, 助教 (10580079)
|
|---|
| Project Period (FY) |
2010 – 2011
|
| Project Status |
Completed (Fiscal Year 2011)
|
| Budget Amount *help |
¥2,717,000 (Direct Cost: ¥2,090,000、Indirect Cost: ¥627,000)
Fiscal Year 2011: ¥1,248,000 (Direct Cost: ¥960,000、Indirect Cost: ¥288,000)
Fiscal Year 2010: ¥1,469,000 (Direct Cost: ¥1,130,000、Indirect Cost: ¥339,000)
|
|---|
| Keywords | 統計的学習理論 / 分類・パタン認識 / 多変量解析 |
|---|
| Research Abstract |
We developed a method to optimize the distribution of features extracted from a given set of data. The features obtained by the method will follow a distribution, which is suitable for classification task. Though the method is applicable to any classification problem, we applied it to speaker recognition task, and achieved state-of-the-art performance. We also developed information estimators, which quantify the information contents in a datum. We applied the estimators to evaluate the credibility of prediction of solar power generated from solar panel.
|
Report
(3 results)
Research Products
(29 results)
