| Budget Amount *help |
¥16,250,000 (Direct Cost: ¥12,500,000、Indirect Cost: ¥3,750,000)
Fiscal Year 2016: ¥5,460,000 (Direct Cost: ¥4,200,000、Indirect Cost: ¥1,260,000)
Fiscal Year 2015: ¥4,940,000 (Direct Cost: ¥3,800,000、Indirect Cost: ¥1,140,000)
Fiscal Year 2014: ¥5,850,000 (Direct Cost: ¥4,500,000、Indirect Cost: ¥1,350,000)
|
|---|
| Outline of Final Research Achievements |
It is often difficult to apply advanced machine learning methods to big data. In such a case, a common approach is to screen out some features and/or instances before the data is fed into machine learning algorithms. Existing screening methods are heuristics in the sense that there is no guarantee that the features and/or instances screened out by the methods are truly irrelevant. In this study, we investigated theory and application of new approach called optimality-guaranteed screening (it is also called safe screening in the literature). We obtained three significant results. The first result is the application of optimality-guaranteed screening to machine learning problems in dynamic environments. The second result is the extension of the scope of optimality-guaranteed screening to the field of pattern mining. The third result is the development of a new method for optimality-guaranteed screening that enables us to screen out features and instances simultaneously.
|
