论文部分内容阅读
随着信息化建设的深入,网络攻击变得复杂多变,严重威胁着网络安全与信息安全.一个好的入侵检测系统往往要求具有高效性,高速性,智能性,实时性,以及应对不同网络环境在线数据的鲁棒性.基于以上五点要求,提出一种权值更新的在线贯序极限学习机算法(WOS-ELM)来应用于网络入侵检测.该算法采用一个一个数据或一块一块数据添加的增量学习算法,将多次迭代求解的神经网络训练转变为一次求解的线性方程组,并通过一种有效的权值赋予的方法来解决网络环境数据不均衡的问题.实验表明,该方法具有很高的正确率,并能在短时间内达到很好的分类效果;较之其他算法,它更适合处理大规模网络实时环境中大量的原始数据,对统计数据依赖性小,对不均衡数据分类具有较好的鲁棒性.因此,基于权值更新的在线贯序极限学习机算法更适应于复杂多变的网络环境下的入侵检测.
With the deepening of information technology, network attacks become complicated and changeable, which seriously threatens network security and information security.A good intrusion detection system often requires high efficiency, high speed, intelligence, real-time, and to deal with different networks Based on the above five requirements, this paper proposes a online weighted sequential learning machine algorithm (WOS-ELM) to apply to network intrusion detection.The algorithm uses one piece of data or piece of data The incremental learning algorithm is added to transform the neural network training which is obtained from iterative iterations into a linear system of equations and solve the problem of unbalanced network environment data by an effective method of weighting.Experiments show that the Compared with other algorithms, it is more suitable for processing a large amount of raw data in large-scale real-time network environment, and has less dependence on statistical data. Balanced data classification has good robustness.Therefore, online sequential limit learning algorithm based on weight updating is more suitable for complex and changeable network environment Detection.