基于混合粒子群优化SVM算法的红斑鳞状皮肤病诊断

DIAGNOSING ERYTHEMATO-SQUAMOUS DISEASE BASED ON HYBRID PARTICLE SWARM OPTIMISATION SVM

红斑鳞状皮肤病的诊断是皮肤病科的一个难题,针对这一问题,提出一种基于混合粒子群的支持向量机(SVM)模型HAPSO-SVM来提高红斑鳞状皮肤病的诊断精度。模型考虑了特征选择机制和参数优化两者对SVM模型起着同等重要的作用,使用自适应的混合粒子群算法(HAPSO)同步实现特征选择机制和SVM的参数寻优,同时设计的线性加权多目标函数综合考虑了分类准确率和支持向量个数,从而提高了算法的准确率和效率。结果表明,提出的模型不仅获得了较少的支持向量个数,找出了红斑鳞状皮肤病紧密相关的特征,并且得到了很高的分类准确率,是一种有效的红斑鳞状皮肤病诊断模型。

The diagnosis of erythemato-squamous disease is a difficult problem in dermatology.In view of this,we propose a hybrid particle swarm optimisation-based support vector machine （SVM）model,namely HAPSO-SVM,for improving the accuracy of erythemato-squamous disease diagnosis.The model takes into account the same important roles on the SVM model played by both the feature selection mechanism and the parameter optimisation,and uses hybrid adaptive particle swarm optimisation （HAPSO）to implement the feature selection mechanism and parameter optimisation simultaneously.Meanwhile,the linear-weighted multi-objective function designed comprehensively considers both the classification accuracy rate and the number of support vectors,therefore improves the accuracy and efficiency of the algorithm.Results show that the proposed algorithm not only achieves small number of support vectors and finds the most related features of erythemato-squamous disease,but also obtains much higher classification accuracy rate,it is proved to be the effective diagnosis model for erythemato-squamous disease.