A COMPARATIVE STUDY OF DEEP LEARNING-BASED MALWARE DETECTION TECHNIQUES FOR ENHANCED NETWORK SECURITY

Network security now faces new and challenging problems as a result of the rise in cyber threats and the creation of malware. Modern security methods are necessary as malicious actors grow more skilled in their techniques. Deep learning-based techniques are better in this aspect since they can learn from vast datasets, spot trends, and achieve a high degree of generalization for unknown forms of malware. This work aims to establish the accuracy of malware detection by proposing and executing a comparative investigation of the different hybrid deep learning frameworks. Metrics like accuracy, precision, recall, scalability, and efficiency used to evaluate these models in relation to how well they recognize known samples and zero-day assaults. To have a thorough and realistic setup, the two malware datasets that are readily available online used for testing as well as training. The study investigates each model's practical applicability in even real-time network security settings, as well as its computing complexity. Therefore, the goal of this research is to further the development of deeper learning-based detection techniques that are more intelligent and adaptable in order to reduce malware threats. In conclusion, the goal of this study is to facilitate the application of theory, allowing for the creation of networks that are more resilient to sophisticated malware attacks in the future. In this research, CNN and ANN models for malware detection. The CNN achieved a high accuracy of 98.44%, while the ANN demonstrated 95.88% accuracy. A comparative analysis of these models was conducted to evaluate their performance, highlighting the strengths and applicability.