In the twenty-first century, when the technology has grown giant, powerful nations learn that supremacy in the cyber world means supremacy in world power. The enormous spying capacity of Pegasus spyware, Natanz nuclear plant attack, cyber-attack in Estonia, manipulation of USA general election through e-mail hacking testifies the above statement. Therefore, defense in the cyber world brings supremacy over any powerful nation. The ability to handle a large amount of data, greater adaptability and state-of-the-art fast detection mechanism has made Artificial Intelligence (AI) very popular in cyber security. Proper training using various datasets, learning through online streaming, and applying multiple learners in cyber defense is practical. This article has been taken to explore cyber defense mechanisms and their efficiency through AI. In doing so, the multiclass portion of the KDD'99 data set has been used. This data set consists of multiple instance types like normal, DoS, U2R, R2L, and Probe. Multiple cyber-attack types are detected under these four broad categories. The data set has been preprocessed exhaustively before starting the testing process in the python lab. Seven ML algorithms are used for detecting multiple cyber-attacks. The high accuracy, the sensitivity of detection by various AI models, and low FPR prove the cyber-defense mechanism's efficiency. Hence, AI is proved to be very efficient in cyber defense mechanisms.
Paper
Full text
State-of-the-Art Artificial Intelligence Based Cyber Defense Model
Semantic Scholar · Computer Science · 2021
Abstract
In the twenty-first century, when the technology has grown giant, powerful nations learn that supremacy in the cyber world means supremacy in world power. The enormous spying capacity of Pegasus spyware, Natanz nuclear plant attack, cyber-attack in Estonia, manipulation of USA general election through e-mail hacking testifies the above statement. Therefore, defense in the cyber world brings supremacy over any powerful nation. The ability to handle a large amount of data, greater adaptability and state-of-the-art fast detection mechanism has made Artificial Intelligence (AI) very popular in cyber security. Proper training using various datasets, learning through online streaming, and applying multiple learners in cyber defense is practical. This article has been taken to explore cyber defense mechanisms and their efficiency through AI. In doing so, the multiclass portion of the KDD'99 data set has been used. This data set consists of multiple instance types like normal, DoS, U2R, R2L, and Probe. Multiple cyber-attack types are detected under these four broad categories. The data set has been preprocessed exhaustively before starting the testing process in the python lab. Seven ML algorithms are used for detecting multiple cyber-attacks. The high accuracy, the sensitivity of detection by various AI models, and low FPR prove the cyber-defense mechanism's efficiency. Hence, AI is proved to be very efficient in cyber defense mechanisms.