Cyber Threat Detection based on Artificial Neural Networks using Event Profile

Abstract

Since cyber threats are getting worse, companies are looking for better ways to analyse security logs and make sure that cyber threats are found quickly and automatically. In this work, we want to use Deep Learning to make a cyber-threat detection framework that is both automated and effective (DL). DL is a promising way to find unknown network intrusions by using self-taught learning. It learns normal and dangerous patterns from the data it collects, taking into account how often they happen and reducing the number of false positive alerts in cyber security. It makes it easier for security analysts to respond quickly to a wide range of cyber threats. PSO was used to improve the accuracy of classification. It ranked all the attributes and chose the features. The SVM algorithm is used to classify the data in these chosen features. The proposed datamodel works well, as shown by the results of experiments on datasets of different sizes.