基于PSO优化孪生支持向量机的自然语言处理  

Natural language processing based on PSO optimized twin support vector machine

作　　者：徐健[1,2] 职燕 刘源[1] Xu Jian;Zhi Yan;Liu Yuan(School of Intelligent Medicine and Biotechnology,Guilin Medical University,Guilin 541199,China;School of Mechanical and Electrical Engineering,Guilin University of Electronic Technology,Guilin 541004,China;College of Mechanical and Control Engineering,Guilin University of Technology,Guilin 541006,China)

机构地区：[1]桂林医学院智能医学与生物技术学院,广西桂林541199 [2]桂林电子科技大学机电工程学院,广西桂林541004 [3]桂林理工大学机械与控制工程学院,广西桂林541006

出　　处：《南京理工大学学报》2023年第1期103-108,共6页Journal of Nanjing University of Science and Technology

基　　金：国家自然科学基金(61474032);桂林医学院博士启动基金(31304019011)。

摘　　要：为了进一步提高孪生支持向量机(Twin support vector machine,TWSVM)的自然语言文本分类准确度,提出了一种改进的粒子群优化(Particle swarm optimization,PSO)算法,并采用改进的PSO算法对TWSVM核心参数进行优化。根据迭代次数来选择自适应权重从而对传统PSO算法进行改进,以防止收敛速度过快而错过全局最优解。采用Word2Vec对自然语言样本进行向量化处理,并通过PSO算法对TWSVM惩罚因子进行优化求解,解决因为惩罚因子设置不合理而造成自然语言文本分类准确率不高的问题。试验证明,通过合理设置PSO算法的速度权重初始值和稳定值,结合自适应递减权重策略,能够获得较高的惩罚因子优化性能,从而提高TWSVM的分类准确率,相比于常见自然语言文本分类算法,PSO-TWSVM的分类准确率更高,均方根误差值更低,在自然语言文本分类中的适用度高。In order to further improve the accuracy of natural language text classification by twin support vector machine(TWSVM),an improved particle swarm optimization(PSO)algorithm is proposed,and the core parameters of TWSVM are optimized by the improved PSO algorithm.Firstly,the traditional PSO algorithm is improved by selecting adaptive weights according to the number of iterations,so as to prevent the convergence speed from being too fast and missing the global optimal solution.Secondly,Word2Vec is used to vectorize natural language samples,and the penalty factor of TWSVM is optimized by PSO algorithm,which solves the problem of low accuracy of natural language text classification caused by unreasonable setting of penalty factor.Experiments show that,by reasonably setting the initial value and stable value of the speed weight of PSO algorithm,combined with the self-adaptive decreasing weight strategy,higher penalty factor optimization performance can be achieved,thus improving the classification accuracy of TWSVM.Compared with common natural language text classification algorithms,PSO-TWSVM has higher classification accuracy,lower root mean square error value and higher applicability in natural language text classification.

关 键 词：自然语言处理 孪生支持向量机 粒子群算法 惩罚因子 自适应权重 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]