摘要
在核极限学习机(Kernel Based Extreme Learning Machine,KELM)分类应用的基础上,结合狮群算法(Loin Swarm Optimization,LSO)强全局寻优能力与收敛快的特性,提出一种LSO优化KELM算法。将测试准确率作为LSO优化KELM的适应度函数,根据移动位置获取最优适应度值进行数据分类测试的评价标准。采用UCI数据集仿真测试,实验结果表明,较KELM分类,LSO优化KELM可获得更优的分类准确率;较麻雀搜索算法(Sparrow Search Algorithm,SSA)优化KELM,LSO优化KELM收敛速度快,分类性能更优。
Based on the classification and application of the kernel based extreme learning machine(KELM),combined with the strong global optimization ability and fast convergence characteristics of the lion swarm optimization(LSO)algorithm,an LSO optimization KELM algorithm is proposed.The test accuracy is taken as the fitness function of LSO to optimize KELM,and the evaluation standard for data classification test is obtained according to the mobile position to obtain the optimal fitness value.Using UCI data set simulation test,the experimental results show that compared with KELM classification,LSO optimization KELM can obtain better classification accuracy.Compared with sparrow search algorithm(SSA)optimization KELM,LSO optimization KELM has faster convergence speed and better classification performance.
作者
刘新建
孙中华
Liu Xinjian;Sun Zhonghua(Wuhan Fiberhome Information Integration Technologies Co.,Ltd.,Wuhan 430074,China)
出处
《电子技术应用》
2022年第2期69-72,共4页
Application of Electronic Technique
关键词
核极限学习机
狮群算法
麻雀搜索算法
kernel extreme learning machine(KELM)
lion swarm optimization(LSO)
sparrow search algorithm(SSA)