摘要
支持向量机是一种新的数据分析方法,应用到越来越多学习问题领域。支持向量机的训练速度和精度与训练集的选取有很大的关系,在对比分析分解算法、SMO(序列最小优化)算法和增量算法的特点和不足的基础上,提出了一种基于Hausdorff距离的训练集的选取方法,利用Libsvm在几个标准数据库上对几种算法进行实验,结果表明,新的算法在速度和精度上具有较大的提高。
Support vector machines is a new method to analyze data and is popular in many learning fields. Its training speed and accuracy depend mainly on the training set selection method. By comparing the characteristic and disadvantage of Chunking, SMO and incremental learning algorithm, a new algorithm is proposed which is based on the Hausdorff distance. Comparing the new algorithm with other methods on some standard databases with the help of Libsvm, and drawing conclusion that the new algorithm can get improvements on accuracy.
出处
《航空兵器》
2007年第5期45-48,共4页
Aero Weaponry