| 构建新包空间的多示例学习方法 |
| |
| 引用本文： | 温超,耿国华,李展.构建新包空间的多示例学习方法[J].西安交通大学学报,2011,45(8):62-66,117. |
| |
| 作者姓名： | 温超 耿国华 李展 |
| |
| 作者单位： | 西北大学信息科学与技术学院,710069,西安 |
| |
| 基金项目： | 国家自然科学基金资助项目(60873094); 高等学校博士学科点专项科研基金资助项目(200806970014) |
| |
| 摘 要： | 针对已有神经网络方法采用示例决定标记从而导致多示例学习(MIL)中包结构信息丢失的问题,提出了一种新的RK_BP多示例学习方法.在示例空间,首先采用粗糙集对其进行属性约简;然后进行K均值聚类,利用聚类点构造新包空间;在新空间中,利用误差反向传播神经网络算法进行分类.在多个测试数据集上对算法进行测试,结果表明该算法可有效解决已有神经网络方法包结构信息丢失问题,明显提高分类性能.
|
| 关 键 词： | 多示例学习 反向传播算法 粗糙集 K均值聚类 新空间 |
Multiple Instance Learning Method Based on Building New Bag Space |
| |
| WEN Chao,GENG Guohua,LI Zhan.Multiple Instance Learning Method Based on Building New Bag Space[J].Journal of Xi'an Jiaotong University,2011,45(8):62-66,117. |
| |
| Authors: | WEN Chao GENG Guohua LI Zhan |
| |
| Institution: | (School of Information Science and Technology,Northwest University,Xi′an 710069,China) |
| |
| Abstract: | Aiming at bag structure information loss problem caused by single instance deciding bag label in multiple instance learning(MIL),a new MIL algorithm named RK_BP is proposed.Firstly,rough set method is adopted to reduce the redundant information in the instance feature space,then K means algorithm is applied to cluster and build a new bag space,and finally back propagation algorithm is used to classify bags in the new space.Experiments on data sets show that this algorithm deals well with the multiple instan... |
| |
| Keywords: | multiple instance learning back propagation rough set K means clustering new space |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
