摘要
多核Fisher判别分析法是一种有效的非线性判别分析法,对其涉及的参数利用遗传算法进行确定是一个有效的途径。针对训练样本较多时遗传算法搜索时间较长的问题,提出一种基于多样本的多核Fisher算法。其做法是将大的训练集拆分成若干个小样本集,依次求得投影映射,并利用"投票策略"来判别待测样本。在人脸识别上的实验表明,基于多样本的多核Fisher算法可以在不降低分类正确率的前提下,提高算法的运算速度。
The multiple kernel Fisher discriminant analysis method is an effective nonlinear algorithm. To determine the related parameters by genetic algorithm is an effective way. The multiple kernel Fisher algorithm based on the diversity samples is presented in this paper to solve the problem that the original algorithm has to spend long time to search when more training samples are adopted. Its approach is that the original training set is splited into a number of small sample sets, which in turn calculate projection mapping by using ＂ voting strategy＂ discrimination for the samples. The experiments for face recognition show that the multiple kernel Fisher algorithm based on the diversity samples can improve the computing speed without reduction of classification correct rate.
出处
《现代电子技术》
2012年第11期73-76,共4页
Modern Electronics Technique
关键词
核方法
多样本
遗传算法
人脸识别
kernel method
diversity sample
genetic algorithm
face recognition
