基于最大—最小蚂蚁系统的动态自适应Ant-Miner算法  

DYNAMIC AND ADAPTIVE ANT-MINER ALGORITHM BASED ON MAX-MIN ANT SYSTEM

作　　者：郭友[1] 黄明和[2] 高山杰[1] 黄超 

机构地区：[1]江西师范大学计算机信息工程学院,江西南昌330022 [2]江西师范大学软件学院,江西南昌330022 [3]胡锦超职业技术学校,广东佛山528305

出　　处：《计算机应用与软件》2012年第9期265-267,共3页Computer Applications and Software

摘　　要：针对基于蚁群算法的文本分类算法Ant-Miner所固有的易陷入局部最优解、复杂度较高等缺陷,引入最大-最小蚁群系统(MMAS)并通过规则质量的变化动态自适应地对信息素的浓度进行调节,引入基于密度的启发式因子来降低计算启发式因子的复杂度,同时通过先验知识来降低条件项选择概率的计算复杂度,在此基础上提出了一种基于Ant-Miner算法的改进文本分类算法。To overcome the inherent defects of ant colony-based text classification algorithm Ant-Miner that it is easy to fall into local optimal solution and is highly complex,we introduce the max-min ant system（MMAS） into the algorithm,and by the variation of rules and quality,the pheromone density is dynamically and adaptively adjusted.The density-based heuristic factor is introduced as well to reduce the complexity of heuristic factor calculation;meanwhile the priori knowledge is employed to reduce the calculation complexity of conditions item selection probability.On these bases we propose an improved Ant-Miner based text classification algorithm.

关 键 词：文本分类 分类规则挖掘 蚁群算法 启发式因子 信息素 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]