Deep Learning For Enhanced Cyber-Attack Detection

With the rapid growth of the Internet of Things, cyber-attacks have also been increasing. To identify and prevent these threats, Intrusion Detection Systems analyze network data traffic patterns. However, processing raw data in an Intrusion Detection System is computationally expensive. This system uses an autoencoder Deep Learning Model to extract features, reduce computational costs, and enhance detection accuracy. Various machine learning classifiers, including K-Nearest Neighbor (K-NN), Random Forest, Gradient Boosting, Logistic Regression, Decision Tree, and Support Vector Machine, are evaluated on autoencoder features. The K-NN classifier outperforms other models with an extraordinary 99.28% accuracy in predicting cyber-attacks. Moreover, the proposed system evaluates the effectiveness of different types of machine learning classifiers in detecting network attacks in the NSL_KDD dataset, while also assessing performance scores for each classifier.