面向机器人视觉识别的近似支持向量机集成方法研究

Research on PSVM Ensemble Algorithm Based on Relief(F) Object to Robot

提出了一种基于Relief(F)的近似支持向量机(PSVM)集成算法.该方法首先在原有的PS-VM集成方法中引入基于Relief(F)算法的特征扰动,以产生具有较大差异性的集成个体,然后应用多数投票法进行集成得到分类结果.

An efficient proximal support vector machine （PSVM） ensemble algorithm based on Relief（F） is proposed in this paper. Firstly,the feature perturbation based on the Relief（F） algorithm is introduced into the original PSVM ensemble method to create the component learners with high diversity. Then the results are gotten by the majority voting.