| 引用本文: | 杨慧中, 王军霞, 丁锋.一种综合信息熵和遗传算法的知识约简方法[J].控制理论与应用,2006,23(6):891~894.[点击复制] |
| YANG Hui-zhong, WANG Jun-xia, DING Feng.Knowledge reduction approach based on information entropy and GA[J].Control Theory and Technology,2006,23(6):891~894.[点击复制] |
|
| 一种综合信息熵和遗传算法的知识约简方法 |
| Knowledge reduction approach based on information entropy and GA |
| 摘要点击 1799 全文点击 1380 投稿时间：2005-04-06 修订日期：2006-03-07 |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| DOI编号 10.7641/j.issn.1000-8152.2006.6.008 |
| 2006,23(6):891-894 |
| 中文关键词 粗糙集 知识约简 信息熵 遗传算法 |
| 英文关键词 rough set knowledge reduction information entropy genetic algorithm |
| 基金项目 国家自然科学基金资助项目(60674092, 60574051). |
|
| 中文摘要 |
| 针对粗糙集理论核心内容之一的知识约简问题, 本文结合信息论有关知识, 给出了粗糙集理论中一些概念和运算的信息表示, 并利用遗传算法作为约简工具, 提出了一种知识相对约简的方法. 为使所得约简相对最优, 将条件信息熵的重要性定义融入了适值函数中. 同时, 在适值函数的选取上引入了惩罚函数和罚系数, 从而保证所求的约简在包含最少而又非零个属性的基础上保持原有的分类能力. 通过实例分析可看出, 该算法对求解约简问题是快速有效的. |
| 英文摘要 |
| One essence of the rough-set theory is knowledge reduction. Computing the minimal knowledge reduction has been proved to be a NP hard problem. Firstly, an information representation of the concepts and operations of rough-set are presented. Secondly, a relative attribute reduction algorithm is developed via the information-entropy-based genetic algorithm (GA). Penalty function and coefficient are then used in fitness function to ensure fewer attributes while keeping consistency of knowledge-base classification in reduction. The significant definition of information entropy used in the fitness function can also make the reduction comparatively optimal. Finally, the experimental result shows that this approach can find the optimal relative attribute reduction effectively and rapidly. |
|
|
|
|
|