Published Paper Details:

Recent advances in computer technology enable the transfer of human existence from actual settings to virtual ones. This development has been hastened by Covid-19 illness. The focus of cybercriminals has also changed from physical to virtual life. This is so because committing a crime online is simpler than it is in real life. Unwanted software known as malicious software (malware) is regularly used by online criminals to launch cyberattacks. Advanced packing and obfuscation methods are being used by malware strains to continue their evolution. These obfuscation methods make malware detection and categorization very difficult. Effectively combating new malware variants requires the employment of novel techniques that are very different from conventional techniques. Deep learning (DL) methods, a type of traditional artificial intelligence (AI), are no longer sufficient to identify all new and sophisticated malware strains. A deep learning (DL) strategy, which differs significantly from conventional DL algorithms, may offer a potential answer to the challenge of identifying all malware types. This work suggests a brand-new deep-learning-based architecture that can categorise malware types using a hybrid approach. The study's key contribution is the suggestion of a novel hybrid design that optimally combines two diverse pre-trained network models. The design of the deep neural network architecture, the construction of the proposed deep neural network architecture, training of the proposed deep neural network architecture, and evaluation of the trained deep neural network comprise the four primary steps of this architecture. On the Malimg, Microsoft BIG 2015, and Malevis datasets, the suggested methodology was tested. The experimental results demonstrate that the proposed method outperforms the state-of-the-art methods described in the literature in its ability to accurately and efficiently categorise malware. On the Malimg dataset, the suggested method demonstrated accuracy of 97.78%, outperforming the majority of DL-based malware detection techniques.