图 书 馆
首页 → 数学
首页 → 物理学
首页 → 物理学
密度偏差抽样技术来提高集群的代表性A Density-Biased Sampling Technique to Improve Cluster Representativeness |
|
| 课程网址: | http://videolectures.net/ecml07_appel_dbs/ |
| 主讲教师: | Ana Paula Appel |
| 开课单位: | 圣保罗大学 |
| 开课时间: | 2008-01-28 |
| 课程语种: | 英语 |
| 中文简介: |  |
| 课程简介: | Data mining algorithms→are computationally very expensive. Sampling→Reduces complexity and Summaries are obtained faster; Clusters have different sizes→problems with uniform sampling; How to sample a multi-dimensional dataset without missing the clusters, even if they are unbalancedand withoutany previous knowledgeabout the clusters? BBS - Biased Box Samplingis a new technique that selects points in such a way to preserve the representativeness of each cluster. It is robust to withstand high-dimensionality drawbacks; insensitive to noise andlinear on both number of pointsNand on the number Eof attributes. |
| 关 键 词: | 密度偏差抽样技术; 提高; 集群的代表性; 数据挖掘算法; 抽样 |
| 课程来源: | 视频讲座网 |
| 最后编审: | 2020-06-15：wuyq |
| 阅读次数: | 68 |