基于神經(jīng)網(wǎng)絡(luò)的入侵檢測(cè)分類算法優(yōu)化設(shè)計(jì)與實(shí)現(xiàn)

1.6萬字
自己原創(chuàng)的畢業(yè)論文，已經(jīng)通過校內(nèi)系統(tǒng)檢測(cè)，重復(fù)率低，僅在本站獨(dú)家出售，大家放心下載使用

摘 要 隨著計(jì)算機(jī)網(wǎng)絡(luò)科技的高速發(fā)展以及廣泛應(yīng)用，網(wǎng)絡(luò)入侵越來越多樣化。傳統(tǒng)的安全技術(shù)所能檢測(cè)到的攻擊十分有限，而入侵檢測(cè)作為一種主動(dòng)防御技術(shù)，它對(duì)進(jìn)出網(wǎng)絡(luò)和主機(jī)的數(shù)據(jù)流進(jìn)行分析判斷，根據(jù)識(shí)別結(jié)果，產(chǎn)生及時(shí)處理。但是單純的入侵檢測(cè)也存在很多缺點(diǎn)，如檢測(cè)海量數(shù)據(jù)效率低、統(tǒng)計(jì)模型較難建立、較高的誤報(bào)率和漏報(bào)率等等，不腀@浞致閿沒棖蟆�
在這種背景下，神經(jīng)網(wǎng)絡(luò)強(qiáng)大的學(xué)習(xí)能力、聯(lián)想記憶能力、模糊運(yùn)算能力和快速尋找優(yōu)化解的能力，使得神經(jīng)網(wǎng)絡(luò)成為解決入侵檢測(cè)系統(tǒng)面臨的問題的一種新途徑。神經(jīng)網(wǎng)絡(luò)的模型非常多，但真正實(shí)用化的并不多，本次畢業(yè)設(shè)計(jì)將選取BP和RBF這兩種應(yīng)用比較廣泛的算法，來實(shí)現(xiàn)高效率、低誤差的入侵檢測(cè)。
本論文會(huì)對(duì)BP和RBF這兩種神經(jīng)網(wǎng)絡(luò)模型的原理進(jìn)行闡述，然后根據(jù)課題要求分別進(jìn)行模塊的設(shè)計(jì)。并且要借助MATLAB平臺(tái)和其中的神經(jīng)網(wǎng)絡(luò)工具箱，利用KDD99數(shù)據(jù)集，兩種不同的運(yùn)行結(jié)果進(jìn)行分析和比較，最終提高入侵檢測(cè)的檢測(cè)率，降低誤報(bào)率。
關(guān)鍵詞：入侵檢測(cè)；神經(jīng)網(wǎng)絡(luò)；BP算法；RBF算法



Optimized design and Implementation of Intrusion detection Classification algorithm based on neural network
Abstract With the rapid development of computer network technology and it’s widely used, network intrusion increasingly diverse. The attack that traditional security technology can detect is very limited. Intrusion detection technology is very proactive; it has access to the network and host data stream analysis to determine. It based on the recognition result, and will have a timely manner. However, the simple IDS has many drawbacks, for example: inefficient detection of massive data, difficult to establish a statistical model, a high rate of false positives and false negatives and so on. It cannot fully meet users’ needs.
In this context, neural network’s powerful learning ability, associative memory capacity, fuzzy computing power and the ability to quickly find the optimal solution, which makes the neural network become a new way to solve the intrusion detection system’s problems. Neural network’s models are too many to choose from, but it’s less to choose something practical. This graduation project will select both BP and RBF algorithm which are widely used, to achieve high efficiency, low error IDS.
In this paper, will the principles of these two algorithms elaborated, with the MATLAB platform and the neural network toolbox in it, using KDD99 data sets, for the results to analyse and compare the advantages and disadvantages. In the end, we can improve the detection rate and reduce the false rate of the intrusion detection.
Key Words：Intrusion detection；Neural network；BP algorithm；RBF algorithm




目 錄
目 錄 III
第一章 緒論 1
1.1 研究背景與意義 1
1.2 國內(nèi)外研究現(xiàn)狀 2
1.3 研究?jī)?nèi)容與工作 2
1.4 論文結(jié)構(gòu)安排 3
第二章 相關(guān)原理與實(shí)驗(yàn)環(huán)境介紹 4
2.1 入侵檢測(cè)簡(jiǎn)介 4
2.1.1 入侵檢測(cè)定義 4
2.1.2 入侵檢測(cè)通用模型 4
2.1.3 入侵檢測(cè)性能指標(biāo) 5
2.2 神經(jīng)網(wǎng)絡(luò)簡(jiǎn)介 6
2.2.1 神經(jīng)網(wǎng)絡(luò)定義 6
2.2.2 神經(jīng)網(wǎng)絡(luò)特點(diǎn) 6
2.2.3 神經(jīng)網(wǎng)絡(luò)模型 7
2.2.4 入侵檢測(cè)基于神經(jīng)網(wǎng)絡(luò)的優(yōu)勢(shì) 9
2.3 實(shí)驗(yàn)環(huán)境介紹 10
2.3.1 MATLAB介紹 10
2.3.2 MATLAB神經(jīng)網(wǎng)絡(luò)工具箱介紹 10
2.3.3 KDD99數(shù)據(jù)集介紹 10
第三章 基于神經(jīng)網(wǎng)絡(luò)的IDS的模塊設(shè)計(jì) 12
3.1 數(shù)據(jù)采集模塊 12
3.1.1 數(shù)據(jù)模塊功能 12
3.1.2 數(shù)據(jù)處理流程 12
3.2 BP模塊設(shè)計(jì) 13
3.2.1 BP算法思想 13
3.2.2 BP算法流程圖 14
3.2.3 BP模型的建立 15
3.3 RBF模塊設(shè)計(jì) 16
3.3.1 RBF算法思想 16
3.3.2 RBF算法流程圖 17
3.3.3 RBF模型的建立 18
第四章 模塊功能實(shí)現(xiàn) 19
4.1 數(shù)據(jù)處理結(jié)果與分析 19
4.2 BP算法運(yùn)行結(jié)果與分析 19
4.3 RBF算法運(yùn)行結(jié)果與分析 23
4.4 兩種算法的對(duì)比 24
第五章 總結(jié)與展望 25
5.1 總結(jié) 25
5.2 展望 25
致 謝 26
參考文獻(xiàn) 27
附錄 代碼 28