摘要
在深度信念网络(Deep Belief Network, DBN)应用于图像、视频等领域中,研究者们普遍通过实践经验设置DBN基本网络结构—DBN深度及每层神经元的个数。将DBN模型作为入侵检测分类模型,提出了DBN模型中基本网络结构的适应度标准,利用该标准提出了一种用于寻找DBN优化网络结构的改进粒子群优化算法。算法首先利用鱼群思想优化粒子群优化算法搜索初始优化网络结构,然后将初始优化网络结构作为利用滑动窗口优化粒子群算法的初始值,继续寻优直到找到全局优化网络结构。将优化算法构造的DBN模型作为入侵检测分类模型进行实验,实验结果表明,相较其它优化算法,方法显著提高了入侵检测分类准确率,明显降低了入侵检测误报率和检测时间,是一种高效且可行的入侵检测分类模型构建和优化方法。
In the field of Deep Belief Network(DBN) applied to images, video, etc., researchers generally set the DBN network structure through experimental experience-DBN depth and the number of neurons per layer. The DBN model is used as the intrusion detection classification model, and the fitness standard of the basic network structure in the DBN model is proposed. We propose an improved particle swarm optimization algorithm for finding the DBN optimized network structure through the standard. The algorithm uses the fish swarm to optimize the particle swarm algorithm to search the initial optimization network structure, and then uses the initial optimization network structure as the initial value of the sliding window optimization particle swarm algorithm, and continues to optimize until it finds a globally optimized network structure. The DBN model constructed by optimization is used as an intrusion detection classification model. The experimental results show that the method improves the classification accuracy of intrusion detection compared with other optimization algorithms, significantly reduces the false alarm rate and detection time of intrusion detection, and proves that it is an efficient and feasible intrusion detection classification model construction and optimization method.
作者
李玉峰
明拓思宇
魏鹏
LI Yu-feng;MING Tuo-si-yu;WEI Peng(School of Computer Engineering and Science,Shanghai University,Shanghai 200444,China;31401 Troops,Jilin City,Jilin Province 132000,China;61660 Troops,Beijing 100089,China)
出处
《计算机仿真》
北大核心
2021年第11期266-274,共9页
Computer Simulation
基金
国家重点研发计划(2017YFB0803201,2016YFB0801200)
国家自然科学基金(61502528)
网络空间安全专项课题(2017YFB0803204)
河南省科技攻关计划课题(162102210034)。
关键词
入侵检测
深度信念网络
人工鱼群算法
粒子群优化
滑动窗口
Intrusion detection
Deep belief network
Artificial fish swarm algorithm
Particle swarm optimization
Sliding window
