题名 |
倒傳遞類神經軟體判別模式 |
并列篇名 |
Process Oriented Event Graph by the Back-Propagation Learning Algorithm |
DOI |
10.30167/JOIT.201012.0021 |
作者 |
邱天嵩(Tien-Sung Chio);林淑民(Shu-Min Lin) |
关键词 |
警訊關聯 ; 事件圖 ; 攻擊圖 ; 倒傳遞類神經網路 ; Alert Correlation ; Event Graph ; Attack Graph ; Back-Propagation Neural Network ; Intelligent |
期刊名称 |
亞東學報 |
卷期/出版年月 |
30期(2010 / 12 / 01) |
页次 |
167 - 184 |
内容语文 |
繁體中文 |
中文摘要 |
隨著1990年代網路基礎建設的普及,使用系統網路上的漏洞來建立攻擊圖以建構資安案件在近年的研究上廣泛被採用,其研究貢獻大多著眼於事件圖的運算效率。本論文提出在現今許多透過合法程式的入侵,在攻擊圖上無法呈現,使得利用攻擊圖所建構的事件關係圖無法精確。根據此點,本研究改以軟體導向事件為基礎來建構事件圖,如此可以避免基於攻擊圖所做的事件關聯中存在許多消失的鏈結。此外,本研究並建立倒傳遞類神經軟體判別模式以輔助不明軟體事件的判定。 |
英文摘要 |
With the network infrastructure being popularly built in the 1990's, utilizing attack graph from system/network exploits to generate the event graph for security scenario construction has been widely adopted by modem research. The contribution concentrates more on the efficiency of computing event graphs. This study intends to address the inconvenience of having attack graph and the inaccuracy of constructing event graph thereafter. We propose correlating the events directly to form the event graph without attack graph and acquiring BP (Back-Propagation) model of Neural Network to assist determining the uncertain events. The other key improvement is to focus more on software oriented event in contrast to network oriented event in the related work. This significantly reduces missing event originally linked to the attack graph based approach. On the whole, the proposed mechanism can improve the accuracy of the event graph and is able to construct a more complete security scenario. |
主题分类 |
人文學 >
人文學綜合 人文學 > 中國文學 基礎與應用科學 > 基礎與應用科學綜合 醫藥衛生 > 醫藥衛生綜合 工程學 > 工程學綜合 社會科學 > 社會科學綜合 |
参考文献 |
|