图 书 馆
perturbo：一种新的基于拉普拉斯Beltrami算子谱的扰动分类算法PerTurbo: a new classification algorithm based on the spectrum perturbations of the Laplace-Beltrami operator |
|
| 课程网址: | http://videolectures.net/ecmlpkdd2011_burger_perturbo/ |
| 主讲教师: | Thomas Burger |
| 开课单位: | 法国南布列塔尼大学 |
| 开课时间: | 2011-10-03 |
| 课程语种: | 英语 |
| 中文简介: |  |
| 课程简介: | Perturbo, an original, nonparametric and effective classification method is proposed here. In our framework, each class of manifold is characterized by its Laplace Beltrami operator, which is evaluated with classical Laplace methods involving graphics. The Classification Benchmark establishes a measure of the size of the spectral disturbance due to this operator. The first experiment shows that the classical algorithms of the most advanced countries have good performance. Moreover, an effective policy design sampling query active learning environment is derived from this measure. Collect toy examples and real-world data sets to evaluate the characteristics of this strategy. |
| 关 键 词: | 非参数; 拉普拉斯; 经典算法 |
| 课程来源: | 视频讲座网 |
| 最后编审: | 2021-08-27：zyk |
| 阅读次数: | 140 |