Deep Learning based Residual Attention Network for Malware Detection in CyberSecurity

As the threat for viruses and malware is increasing so detecting them accurately is very important. By studying and researching different papers, it is finally known that the main problem about malware detection has recorded high false positive rates and malware obfuscation. To deal with this limitation, this research work has created a model for testing the dataset based on attention neural network as it is the latest topic of research. The proposed Residual Neural Network uses an extended non-linear kernel and composite skip connections to improve the model’s ability to classify and detect hidden malware patterns. With the new model based on attention network, it is observed that by paying attention to specific features the performance of detection can be improved and the severity of damage done by malwares or viruses can be reduced. The proposed cybersecurity model is also compared with conventional machine learning and deep learning techniques.