基于改进DS证据融合与ELM的入侵检测算法

发布时间：2018-06-05 06:57

  本文选题：网络入侵检测 + DS证据理论　； 参考：《计算机应用研究》2016年10期

【摘要】：为了提高检测率,采用DS证据融合技术融合多个ELM,能够提高整个检测系统的精确性。但是传统的DS技术处理冲突信息源时并不理想。因此,通过引入证据之间的冲突强度,将信息源划分成可接受冲突和不可接受冲突,给出了新的证据理论(improved Dempster-Shafer,I-DS),同时针对ELM随机产生隐层神经元对算法性能造成影响的缺陷作出了改进。通过实验表明,结合I-DS和改进的ELM能够更高速、更有效地判别入侵行为。
[Abstract]:In order to improve the detection rate, the accuracy of the whole detection system can be improved by using DS evidence fusion technology to fuse multiple ELMs. However, the traditional DS technology is not ideal in dealing with conflict information sources. Therefore, by introducing the intensity of conflict between evidence, the information source is divided into acceptable conflict and unacceptable conflict. In this paper, a new evidence theory is presented, and an improvement is made to the defect that the random generation of hidden layer neurons in ELM has an effect on the performance of the algorithm. Experiments show that the combination of I-DS and improved ELM can distinguish intrusion behavior more efficiently and efficiently.
【作者单位】： 江苏科技大学计算机科学与工程学院;
【基金】：国家自然科学基金资助项目(61305058) 江苏省自然科学基金资助项目(BK20130471)
【分类号】：TP393.08
，

本文编号：1981065