Computer Engineering and Applications ›› 2017, Vol. 53 ›› Issue (9): 158-161.DOI: 10.3778/j.issn.1002-8331.1511-0121

Previous Articles     Next Articles

Improved chicken swarm optimization and its application in coefficients optimization of multi-classifier

HONG Yang, YU Fengqin   

  1. School of Internet of Things Engineering, Jiangnan University, Wuxi, Jiangsu 214122, China
  • Online:2017-05-01 Published:2017-05-15

改进的鸡群算法并用于多分类器系数优化

洪  杨,于凤芹   

  1. 江南大学 物联网工程学院,江苏 无锡 214122

Abstract: For CSO can not get the global optimal value due to the roosters are strapped into??local minima, an improved CSO based on hybrid mutation mechanism is proposed, which applies the hybrid mutation after the hens are updated, making the hens speed up leaving the local optimization points; at the same time, the roosters are replaced by hens with good fitness through the updating mechanism, then the coefficients of the multi-classifier are optimized by the algorithm. The experimental results show that the improved CSO can effectively avoid the local minima. The multi-classifier optimized by the improved CSO has lower error rate and training time.

Key words: improved Chicken Swarm Optimization(CSO), hybrid mutation, multi-classifier, coefficients optimization

摘要: 针对鸡群算法因雄鸡粒子易陷入局部最优而无法取得全局最优问题,提出了基于杂交的改进的鸡群算法。即在雌鸡粒子更新后加入杂交机制,使雌鸡粒子加速离开局部最优点;同时通过更新机制将性能优越的雌鸡粒子设定为雄鸡粒子来避免雄鸡粒子陷入局部最优,并将改进的鸡群算法用于多分类器系数的优化。仿真实验结果表明,改进的鸡群算法不易陷入局部最优,且用该算法优化的多分类器其错误率降低,训练时间缩短。

关键词: 鸡群算法, 杂交, 多分类器, 系数优化