Using Density Estimation to Improve the Learning with Small Data Sets
碩士 === 國立成功大學 === 工業管理科學系碩博士班 === 91 === This study is devoted to learn knowledge with a small data set using statistical learning theory. Since fewer exemplars usually lead to a lower learning accuracy, many approaches use a big number of exemplars in learning process for higher learning accuracies...
| Main Authors: | , |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en_US |
| Published: |
2003
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/28652252338863398522 |
| Summary: | 碩士 === 國立成功大學 === 工業管理科學系碩博士班 === 91 === This study is devoted to learn knowledge with a small data set using statistical learning theory. Since fewer exemplars usually lead to a lower learning accuracy, many approaches use a big number of exemplars in learning process for higher learning accuracies. However, the idea would be inappropriate when a research is limited by cost and time. To overcome this difficulty, this research uses kernel methods of Density Estimation to improve the small size learning. Furthermore, Virtual Samples Generation with Intervalized Kernel Density Estimation is proposed to produce enough information for learning. The provided example shows that this is an economical and efficient method of knowledge acquisition from small data sets.
|
|---|