Intelligent Malware Detection: Enhancing Accuracy with Recurrent Neural Network

Aim: The purpose of this research is to use Artificial Neural Networks (ANN) to create an intelligent malware detection system and evaluate how well it performs in comparison to ClamA V, a conventional antivirus program that relies on signatures. By using Recurrent Neural Networks (RNN) for feature extraction and classification, the study aims to increase detection accuracy. Materials and Methods: Two groups are taken into consideration in this study. Group 1 is the ANN-based intelligent malware detection system, ClamA V which uses with 26 samples. Group 2 is the 26- sample RNN signature-based detection algorithm. With a 95% confidence interval and a threshold of 0.05%, the G Power value is 80%. Both static and dynamic features taken from malware samples are processed by the suggested RNN -based malware detection method. The collection includes labelled To increase convergence, a backpropagation technique with an adaptable learning rate is used to train the ANN model. Results: The ANN- based intelligent malware detection system outperforms the ClamA V system in terms of detection accuracy (0.92 vs. 0.85), precision (0.88 vs. 0.80), recall (0.91 vs. 0.63), and F1-score (0.91 vs. 0.81). While the ClamA V detection accuracy ranges from 55.23% to 78.650/0, the RNN-based system's detection accuracy ranges from 85.42% to 98.76%. At a significance level of 0.0047 and a learning rate of 0.001, the best classification accuracy is obtained. Conclusion: This study shows that the suggested ANN-based malware detection system more especially, the RNN performs noticeably better than the conventional ClamA V system.