Algorithmic Strategies for Cyber Crime Attack Prevention Harnessing the Power of Convolutional Neural Networks
J. Christina Deva Kirubai, S. Silvia Priscila · 2024
Cybercrime is a major concern in today's cyber world, and immediate action must be taken to prevent it. Traditional cybersecurity systems are vulnerable to acclimatization issues, a high percentage of false positive alerts, and so forth. To address these problems, the study proposes a cutting-edge technique that blends CNNs with transmission characteristics for anti-cybercrime policy. The CN-based system is designed to retrieve network traffic data in real-time, with the reliability of identified hostile activity proportional to the minimizing of false alarms. The system includes components such as data collection and preprocessing, CNN architecture design, training, real-time analysis, and adaptation to emerging threats. Superior performance characteristics can be seen in comparison to existing systems, including 0.90-0.92 greater accuracy, 0.89-0.94 precision, 0.97-0.98 recall, and 0.98-0.93 F1-score with a ROC AUC score of 0.97. Furthermore, the proposed system has faster detection speeds (1200 packets per second) and lower false positive rates for various cyber threat types such as malware (0.03), phishing assaults (0.05), and DDoS attacks (0.02). Thus, the proposed system represents a big step forward in the evolution of cyber security, adopting more active threat identification and mitigation, thereby strengthening existing network security and exposing it to fewer threats.