欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
针对入侵检测中训练样本数量多、属性多这一问题，应用核主成份分析KernelPCA和简约支持向量机Reduced SVM相结合的方法，不但有效地提取了样本的非线性信息，而且使样本在维数上得到约简，减少了核矩阵的计算量。在标准入侵检测数据集上的实验表明：训练时间进一步减少，正确率得到提高，而误报率下降。
英文摘要：
The massive problems in the intrusion detection induced by too many instances and attributes are studied. A new method using Kernel PCA and Reduced SVM is proposed. The method not only extracts the nonlinear information from the samples effectively but also reduces dimensions and the computation requirements of the kernel matrix. Experiment implemented on the standard dataset shows that it needs little training time and the accuracy is higher while the false positive rate is lower.
同期刊论文项目
同项目期刊论文
期刊信息
位置：