摘要
支持向量机一直被用于人脸识别,为了提高其分类识别的精度,文中提出一种改进的麻雀搜索算法优化支持向量机参数的分类模型。首先,利用Tent混沌映射有效地改善了由于初始种群缺乏遍历性和多样性差而导致的收敛速度慢的问题;其次,借鉴粒子群算法的学习策略和云模型引入自适应动态权重因子、平衡全局和局部搜索能力,扩大算法的搜索范围;最后,引入Levy飞行,提高算法寻优能力和跳出局部极值的能力。从基准测试函数结果不难看出,ISSA比SSA、PSO、BOA和GA算法收敛速度更快,搜索能力更准确,更容易跳出局部极值。基于LFW人脸数据集实验结果表明,所提方法的平均识别率为89.36%,优于其他方法,验证了其有效性。
Support vector machine has been used for face recognition,in order to improve the accuracy of its classification recognition,this paper proposes an improved sparrow search algorithm to optimize the classification model of support vector machine parameters.Firstly,the use of Tent chaos mapping effectively improves the slow convergence speed caused by the lack of ergocity and poor diversity of the initial population.Secondly,the adaptive dynamic weight factor is introduced by drawing on the learning strategy of particle swarm algorithm and cloud model,balancing global and local search capabilities,and expanding the search scope of the algorithm.Finally,Levy flight is introduced to improve the algorithm's optimization ability and ability to jump out of local extremes.It is not difficult to see from the results of the benchmark function that ISSA converges faster,has more accurate search capabilities,and is easier to jump out of local extremes than the five algorithms of SSA,PSO,BOA and GA.The experimental results based on LFW face dataset show that the average recognition rate of the proposed method is 89.36%,which is better than other methods,and its effectiveness is verified.
作者
赵明明
胡茜
赵婧帆
ZHAO Mingming;HU Qian;ZHAO Jingfan(School of Mathematics&Statistics,Changchun University of Technology,Changchun 130012,China)
出处
《长春工业大学学报》
CAS
2024年第5期472-480,共9页
Journal of Changchun University of Technology
基金
吉林省教育厅社科项目(JKH20220649SK)。
关键词
麻雀搜索算法
支持向量机
人脸识别
SSA(Sparrow Search Algorithm)
SVM(Support Vector Machine)
face recognition.
