支持向量机是当前智能计算研究领域的热点之一。基于支持向量机的大样本学习一直是一个非常具有挑战性的研究课题。对于分类问题给出一种基于相似度的约简数据集的方法。给出的新算法大大地减少了训练样本的数目和所求解的支持向量机算法的规模,有效地加快了支持向量机算法的训练速度。仿真实验表明:新算法较为简单、实用。
Support Vector Machine(SVM) has become a hot spot in the field of intelligence computing. Also,the learning of big sample based on SVM has been being a challenging task. A new reduced set method based on similarity is presented for classification. The quantity of the training sets and the scale of the SVM algorithm are reduced greatly, and the training of the SVM is accelerated. The simulation result demonstrates the feasibility of the new algorithm.