基于分布式學習的神經(jīng)網(wǎng)絡入侵檢測算法研究
發(fā)布時間：2018-12-15 12:45
【摘要】：當今社會，計算機網(wǎng)絡發(fā)展迅速，確保網(wǎng)絡信息的安全性就顯得尤為重要。能夠主動保護信息安全的入侵檢測技術，作為一種保障措施而備受關注。神經(jīng)網(wǎng)絡的優(yōu)勢在于，它能夠作為一種方法應用到入侵檢測中。通過分析和訓練大量的實例數(shù)據(jù)，神經(jīng)網(wǎng)絡學習訓練的知識，根據(jù)已有的實例，自主掌握并分析出系統(tǒng)中各個實例和變量之間的關系，而不需要了解數(shù)據(jù)分布和解析的細節(jié)。 本文主要對入侵檢測的概念、功能以及檢測方法進行詳細的介紹，并詳細闡述了神經(jīng)網(wǎng)絡的概念、工作原理以及神經(jīng)網(wǎng)絡的研究內(nèi)容。重點介紹了BP算法的原理、步驟以及流程，根據(jù)BP神經(jīng)網(wǎng)絡模型的特點，通過比較算法的優(yōu)缺點，對現(xiàn)有算法進行改進。 首先，從神經(jīng)網(wǎng)絡的原理入手，對該原理進行討論，研究了傳統(tǒng)BP網(wǎng)絡學習算法，并結合分布式和自適應的特點，對傳統(tǒng)BP算法進行改進，提出了一種優(yōu)化的BP神經(jīng)網(wǎng)絡入侵檢測算法，即分布式神經(jīng)網(wǎng)絡自學習算法。通過改進的算法對入侵數(shù)據(jù)進行檢測和學習，直接使用BP學習方法的訓練樣本數(shù)量過大而且不易收斂，這一問題得到了很好地解決。 其次，通過對改進算法的研究，給出算法的具體步驟，并運用改進的算法來建立模型，對該模型進行分析，與傳統(tǒng)BP網(wǎng)絡學習算法進行對比，，驗證改進算法的可行性與有效性。 最后，將算法應用于入侵檢測中，通過相應的測試方法，對本文所采用的樣本數(shù)據(jù)集來進行實例驗證。通過檢測數(shù)據(jù)的測試結果，驗證分布式神經(jīng)網(wǎng)絡自學習算法的性能，得出結論。
[Abstract]:Nowadays, with the rapid development of computer network, it is very important to ensure the security of network information. Intrusion detection technology, which can actively protect information security, has attracted much attention as a safeguard. The advantage of neural network is that it can be applied to intrusion detection as a method. By analyzing and training a large number of case data, neural networks learn the knowledge of training, according to the existing examples, independently grasp and analyze the relationship between each instance and variables in the system, without the need to understand the details of data distribution and analysis. In this paper, the concept, function and detection method of intrusion detection are introduced in detail, and the concept, working principle and research content of neural network are described in detail. The principle, steps and flow of BP algorithm are introduced emphatically. According to the characteristics of BP neural network model, the advantages and disadvantages of the algorithm are compared and the existing algorithms are improved. Firstly, this paper discusses the principle of neural network, studies the traditional learning algorithm of BP network, and combines the characteristics of distributed and adaptive, improves the traditional BP algorithm. An optimized BP neural network intrusion detection algorithm, distributed neural network self-learning algorithm, is proposed. By using the improved algorithm to detect and learn intrusion data, the number of training samples using BP learning method is too large and difficult to converge. This problem has been solved well. Secondly, through the research of the improved algorithm, the concrete steps of the algorithm are given, and the model is established by using the improved algorithm, and the model is analyzed, and compared with the traditional BP network learning algorithm, the feasibility and effectiveness of the improved algorithm are verified. Finally, the algorithm is applied to intrusion detection, and the sample data set used in this paper is verified by the corresponding test method. The performance of the distributed neural network self-learning algorithm is verified by the test results of the data, and the conclusion is drawn.
【學位授予單位】：哈爾濱理工大學
【學位級別】：碩士
【學位授予年份】：2014
【分類號】：TP393.08;TP183
本文編號：2380651
[Abstract]:Nowadays, with the rapid development of computer network, it is very important to ensure the security of network information. Intrusion detection technology, which can actively protect information security, has attracted much attention as a safeguard. The advantage of neural network is that it can be applied to intrusion detection as a method. By analyzing and training a large number of case data, neural networks learn the knowledge of training, according to the existing examples, independently grasp and analyze the relationship between each instance and variables in the system, without the need to understand the details of data distribution and analysis. In this paper, the concept, function and detection method of intrusion detection are introduced in detail, and the concept, working principle and research content of neural network are described in detail. The principle, steps and flow of BP algorithm are introduced emphatically. According to the characteristics of BP neural network model, the advantages and disadvantages of the algorithm are compared and the existing algorithms are improved. Firstly, this paper discusses the principle of neural network, studies the traditional learning algorithm of BP network, and combines the characteristics of distributed and adaptive, improves the traditional BP algorithm. An optimized BP neural network intrusion detection algorithm, distributed neural network self-learning algorithm, is proposed. By using the improved algorithm to detect and learn intrusion data, the number of training samples using BP learning method is too large and difficult to converge. This problem has been solved well. Secondly, through the research of the improved algorithm, the concrete steps of the algorithm are given, and the model is established by using the improved algorithm, and the model is analyzed, and compared with the traditional BP network learning algorithm, the feasibility and effectiveness of the improved algorithm are verified. Finally, the algorithm is applied to intrusion detection, and the sample data set used in this paper is verified by the corresponding test method. The performance of the distributed neural network self-learning algorithm is verified by the test results of the data, and the conclusion is drawn.
【學位授予單位】：哈爾濱理工大學
【學位級別】：碩士
【學位授予年份】：2014
【分類號】：TP393.08;TP183
【引證文獻】
相關碩士學位論文 前2條
1 趙菁偉;基于分簇Ad Hoc網(wǎng)絡的入侵檢測系統(tǒng)設計[D];河北科技大學;2016年
2 許鋒;人工神經(jīng)網(wǎng)絡與遺傳算法相結合的入侵檢測模型的研究[D];江蘇科技大學;2015年
本文編號：2380651
本文鏈接：http://sikaile.net/guanlilunwen/ydhl/2380651.html
最近更新
教材專著
