Journal of Frontiers of Computer Science and Technology ›› 2016, Vol. 10 ›› Issue (5): 722-731.DOI: 10.3778/j.issn.1673-9418.1506093

Previous Articles     Next Articles

Ant Lion Optimizer with Chaotic Investigation Mechanism for Optimizing SVM Parameters

ZHAO Shijie+, GAO Leifu, YU Dongmei, TU Jun   

  1. Institute of Optimization and Decision, Liaoning Technical University, Fuxin, Liaoning 123000, China
  • Online:2016-05-01 Published:2016-05-04

带混沌侦查机制的蚁狮优化算法优化SVM参数

赵世杰+,高雷阜,于冬梅,徒  君   

  1. 辽宁工程技术大学 优化与决策研究所,辽宁 阜新 123000

Abstract: As ant lion optimizer (ALO) is a new bionic intelligence algorithm, there are a number of respects on the improvement and development. Since antlion’s population (species) has some poor-fitness individuals in basic ALO algorithm, the behavior of ants selecting those antlions for random walk will result in increasing the possibility of its trapping into local optima and impacting on the algorithm’s optimal performance. Considering this question, this paper proposes ant lion optimizer with chaotic investigation mechanism (CIALO), which draws experience from the investigation idea of artificial bee colony algorithm (ABC) and brings in chaos search mechanism based on the original information of antlions. The CIALO algorithm firstly defines poor-fitness individuals of the sorted antlions’ population as investigative ant lions (IAL). Meanwhile, the original position information of these antlions is regarded as the initial value of Fuch chaotic mapping. Then it can gain a better-much position by a certain number of chaos search iteration and reassigns the position to IAL, which is beneficial to improve the superiority of antlion’s population and the optimal performance of the algorithm. Eventually, the CIALO algorithm is used to optimize the parameters of support vector machine (SVM). The public datasets from University of California Irvine (UCI) is employed for evaluating the proffered algorithm. The experimental results imply that the CIALO algorithm for optimizing SVM parameters has stronger optimal performance and better stability of the algorithm.

Key words: ant lion optimizer, chaos, investigation mechanism, support vector machine, parameter optimization

摘要: 蚁狮优化算法作为一种新的仿生智能算法,有许多有待完善和发展的方面。由于在算法迭代过程中蚁狮种群存在适应度相对较差的个体,若蚂蚁选定该蚁狮进行随机游走将会增加算法陷入局部极值的可能性,同时会影响算法的寻优性能。针对该问题,借鉴人工蜂群算法的侦查思想,在蚁狮原有信息的基础上引进混沌搜索机制,提出了一种带混沌侦查机制的蚁狮优化算法。该算法首先将排序蚁狮种群中适应度较差的个体定义为侦查蚁狮,并将其原始位置信息作为Fuch混沌映射的初始值,然后通过一定次数的混沌搜索迭代获得一个适应度值更优的位置再重新赋值给侦查蚁狮,以提高蚁狮种群的优良性和算法的寻优性能。最后将改进蚁狮优化算法用于支持向量机参数的优化中,以UCI标准数据库中的数据进行数值实验,结果表明改进算法具有较强的寻优性能和较好的算法稳定性。

关键词: 蚁狮优化算法, 混沌, 侦查机制, 支持向量机, 参数优化