基于IPSO算法的特征选择与目标识别  

Feature selection and target recognition based on improved particle swarm optimization algorithm

作　　者：吴莉[1] 梅雪[1] 林锦国[1] 

机构地区：[1]南京工业大学自动化与电气工程学院,江苏南京210009

出　　处：《计算机工程与设计》2011年第11期3832-3835,共4页Computer Engineering and Design

基　　金：江苏省高校自然科学研究基金项目(09KJB510002);江苏省博士后科研计划基金项目(1001027B);南京工业大学青年学术基金项目(39710006)

摘　　要：为了提高相似目标的分类识别率,实现降维,提出了一种基于改进的粒子群优化(IPSO)的特征选择与目标识别方法。IPSO利用二进制位串来计算位置和速度,并在速度更新公式中增加约束项,权衡识别率与特征维数的比重选择适应度函数。结合距离分类器,用IPSO在自建的相似目标特征库上进行最优特征子集选择及分类实验。实验结果表明了该算法的有效性,在UCI数据集上的对比实验结果表明了IPSO的改进效果。A method for feature selection and target recognition based on improved particle swarm optimization （IPSO） algorithm is proposed for improving classification accuracy and reducing feature dimensions of similar targets. In IPSO algorithm, binary bit string is used to calculate the position, velocity and a constraint item is added in the velocity updating formula and the proportion of recognition rate and feature dimension is weighted to choose the fitness function. IPSO is used to select the optimum feature subset and distance classifier is used to perform the classification on the own-built datasets of similar target. Experimental results show the effectiveness of the new method. Besides, comparing experiment on standard datasets from UCI machine learning is finished to verify the improvement effectiveness of IPSO algorithm.

关 键 词：改进粒子群优化(IPSO) 二进制粒子群优化(BPSO) 特征选择 小波不变矩 相似目标 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]