| 摘要: |
| 为了提高球形分类器的分类性能,受支持向量机和小球体大间隔等方法的启发,提出一种大间隔最小压缩包含球(large margin and minimal reduced enclosing ball,简称LMMREB)学习机,其在Mercer 核诱导的特征空间,通过优化一个最小包含球,以寻求两个同心的分别包含二类模式的压缩包含球,且使二类模式分别与压缩包含球间最小间隔最大化,从而可以同时实现类间间隔和类内内聚性的最大化.分别采用人工数据和实际数据进行实验,结果显示, LMMREB 的分类性能优于或等同于相关方法. |
| 关键词: 泛化 支持向量数据描述 支持向量机 最小包含超球体 |
| DOI：10.3724/SP.J.1001.2012.04071 |
| 分类号: |
| 基金项目:国家自然科学基金(60975027, 60903100); 宁波市自然科学基金(2009A610080) |
|
| Large Margin and Minimal Reduced Enclosing Ball Learning Machine |
|
TAO Jian-Wen1,2, WANG Shi-Tong1
|
|
1.School of Information Engineering, Jiangnan University, Wuxi 214122, China;2.School of Information Engineering, Zhejiang Business Technology Institute, Ningbo 315012, China
|
| Abstract: |
| In this paper, inspired by the support vector machines for classification and the small sphere and large margin method, the study presents a novel large margin minimal reduced enclosing ball learning machine (LMMREB) for pattern classification to improve the classification performance of gap-tolerant classifiers by constructing a minimal enclosing hypersphere separating data with the maximum margin and minimum enclosing volume in the Mercer induced feature space. The basic idea is to find two optimal minimal reduced enclosing balls by adjusting a reduced factor parameter q such that each of binary classes is enclosed by them respectively and the margin between one class pattern and the reduced enclosing ball is maximized. Thus the idea implements implementing both maximum between-class margin and minimum within-class volume. Experimental results obtained with synthetic and real data show that the proposed algorithms are effective and competitive to other related diagrams. |
| Key words: generalization support vector data description support vector machine minimum enclosing hypersphere |
