Abstract: This paper presents two Support Vector Machine（SVM） training algorithms based on Adaptive Resonance Theory（ART） and Self-Organizing feature Map（SOM） neural networks，namely ART-SVM algorithm and SOM-SVM algorithm respectively，in order to improve learning efficiency of SVM on large scale datasets.By clustering the original data，the given data can be reduced greatly.In so doing，the speed of SVM training can be greatly improved and the satisfactory generalization performance can be obtained as well.

Key words: Support Vector Machine（SVM）, ART-SVM algorithm, SOM-SVM algorithm, cluster

摘要： 针对支持向量机（Support Vector Machine，SVM）对大规模样本分类效率低下的问题，提出了基于自适应共振理论（Adaptive Resonance Theory，ART）神经网络与自组织特征映射（Self-Organizing feature Map，SOM）神经网络的SVM训练算法，分别称为ART-SVM算法与SOM-SVM算法。这两种算法通过聚类压缩数据集，使SVM训练的速度大大提高，同时可获得令人满意的泛化能力。

关键词: 支持向量机, ART-SVM算法, SOM-SVM算法, 聚类