DEEP ENSEMBLE-BASED EFFICIENT FRAMEWORK FOR NETWORK ATTACK DETECTION

Abstract

Networks are essential for numerous sports, including corporate operations, instructional hobbies, and each day long-distance conversation. Although networks provide several advantages, additionally they pose safety risks that can jeopardise statistics confidentiality, integrity, and privacy. Network threats, including virus, hacking, and phishing, are increasing, leading to enormous economic and reputational harm. The assignment proposes the introduction of an automatic system utilizing artificial intelligence (AI) to mitigate those protection vulnerabilities. This generation is designed to effectively become aware of and protect in opposition to community threats, for this reason improving the security of statistics and networked structures. The mission implements an ensemble model that integrates 3 deep learning architectures: LSTM, RNN, and GRU. These models collaborate employing majority balloting standards to get extended accuracy inside the identification of network attacks, therefore making sure strong safety for networked environments. The venture augments its ability through the incorporation of a “Voting Classifier (Random Forest AdaBoost) and a Stacking Classifier”, the latter achieving an amazing a 100% accuracy, which demonstrates improved effectiveness in detecting community attacks.