An Efficient Network Attack Detection System Using Recurrent Neural Network Models
Amogh Muragodmath, Arfaahussain Shaikh, Naveenkumar Baraker, Deepak Baligar, Priyadarshini C. Patil · 2024
This study addresses the escalating complexity of cyber threats through the application of deep learning models and time series visualization, aiming to enhance the efficiency of network attack detection. To establish robust defense mechanisms, we develop a predictive framework utilizing neural networks. This framework incorporates a unique structure that synergizes long short-term memory (LSTM) and recurrent neural networks (RNN s) to effectively handle sequential data, leading to the identification of potential threats. Diverse datasets, encompassing network traffic patterns, system vulnerabilities, and historical attack data, are employed to train the models. The developed framework achieves an impressive 99 % accuracy in attack detection, emphasizing its effectiveness in ensuring cybersecurity. Furthermore, our approach integrates model predictions with visual representations of network activity, augmenting overall de-tection accuracy. Performance evaluations are conducted across various attack scenarios and network configurations, contributing to a comprehensive understanding of the model's capabilities and practical implications.