Optimizing Real Time Intrusion Detection for Enhanced Network Security
Namala Madhuri, R. Tamilkodi, Katari Meghana, G. Amrutha, Sada Surendra, Megha Hari · 2025
This addresses the growing threats of computer viruses, malware, and cyberattacks on networks, emphasizing the importance of intrusion detection as a proactive defense mechanism. Existing solutions primarily rely on DNN, which, while effective, face challenges in accuracy and false positive rates. To enhance cybersecurity, we propose innovative algorithms, including CNN, RNN with LSTM, DNN, RBM combined with BILGRU, and a hybrid CNN+LSTM model. Our hybrid approach achieved an impressive 99% accuracy on the KDD-Cup dataset, significantly improving detection effectiveness and reducing false positives. Utilizing a generative adversarial network and contrasting unaided and DL-based discriminative techniques, we offer a total methodology that further develops ID in digital actual frameworks controlled by the IoT. This work addresses a significant accomplishment nearby and adds to more vigorous protection against changing digital assaults by featuring the capability of complex DL procedures in handling network safety shortcomings..