摘要
群智能算法中的粒子群算法通过模仿鸟群的觅食过程,进行迭代寻求最优解值,被广泛应用于机器学习分类算法改进与参数寻优。为进一步提高随机森林模型分类准确度,提出利用柯西变异改进粒子群算法,将柯西扰动项加入粒子属性变动过程,并将已改进的粒子群算法用于随机森林模型中的参数优化过程,构建随机森林模型C_SO_F。在UCI数据集上通过独立重复对比实验将C_SO_F与经典随机森林模型进行研究与数据分析,结果表明,结合柯西变异的粒子群算法提高了随机森林模型的分类准确度与模型稳定性。
The particle swarm algorithm in swarm intelligence algorithm seeks the optimal solution value by imitating the foraging process of a flock of birds,and is widely used in machine learning classification algorithm improvement and parameter search. In order to further improve the classification accuracy of the random forest model,we propose to improve the particle swarm algorithm by adding the Cauchy variation term to the particle property change process,and use the improved particle swarm algorithm for the parameter optimization process in the random forest model to construct the random forest model C_SO_F. In UCI dataset,C_SO_F and classical random forest model is studied and analyzed by independent repetitive comparison experiments. The results show that the particle swarm algorithm combined with the Cauchy variation improves the classification accuracy and model stability of the random forest model.
作者
赵成兵
刘慧慧
谢新平
刘静
ZHAO Chengbing;LIU Huihui;XIE Xinping;LIU Jing(School of Mathematics and Physics,Anhui Jianzhu University,Hefei 230022,China)
出处
《宿州学院学报》
2021年第12期9-12,52,共5页
Journal of Suzhou University
基金
安徽省高等学校自然科学基金重点项目(KJ2020A0479,KJ2018A0517)。
关键词
粒子群算法
柯西变异
随机森林
机器学习
集成算法
Particle swarm optimization
Cauchy variation
Random forest
Machine learning
Integrated algorithm
