摘要
本文提出了基于RBF-HMM模型的网络入侵检测方法,给出了该模型的训练和识别方法。因为HMM模型的分类决策能力和对不确定信息的描述能力不理想,而人工神经网络对动态时间序列的建模能力尚不尽如人意,所以将RBF神经网络集成到HMM框架中,用RBF神经网络为HMM提供状态概率输出。通过RBF神经网络的粗分类,克服了HMM的缺陷,提高了系统的分类识别能力。
A network intrusion detection framework and its associated algorithm based on RBF-HMM are put forward,the training and identification methods of the algorithm are given.HMM model for the classification decision-making capacity and ability to describe the uncertain information is not ideal,and Artificial Neural Networks for dynamic time series modeling capabilities is not yet satisfactory,so the RBF neural network is used to provide state probability output forHMM in the HMM framework.With the rough sort of RBF,the limitation of HMM is overcome;the ability of classification and identification is enhanced.
出处
《网络安全技术与应用》
2011年第1期9-11,共3页
Network Security Technology & Application
基金
湖南省学位与研究生教育教学改革研究课题(项目编号:08B38)资助
关键词
RBF-HMM
向量量化
前(后)向评估算法
入侵检测
Radial Basis Function-hidden Markov model
Vector Quantification
Forward（backward）evaluation algorithm
intrusion detection
