支持向量机在大规模数据分类中的应用

Application of Support Vector Machines in Large-scale Data Classification

笔者对支持向量机在大规模数据分类中的应用进行了分析研究,针对其在处理大规模数据时存在的内存空间需求过大和训练时间过长的问题,基于不同特性的通用数据集对三种不同算法支持向量机的分类时间和准确度进行了实验,比较分析了传统支持向量机采用不同核函数和线性支持向量机采用不同损失函数的分类效果,对分布式支持向量机的分类进行了实验。实验结果表明,相对于传统支持向量机,分布式支持向量机和线性支持向量机的训练速度和分类准确率都有提高。

The author analyzes the application of SVM in large-scale data classification.According to the problem of large memory space requirement and long training time,aiming at the problem that three kinds of common data set based on different characteristics,experiments were carried out on the classification time and accuracy of different SVMs.The classification results of traditional SVM with different kernel functions and linear support vector machines using different loss functions were compared and analyzed.The classification of distributed SVM was tested.The experimental results show that the training speed and classification accuracy of distribute SVM and linear SVM are improved compared with traditional SVM.