Advance Malware Variant Detection Utilizing CRNN Hybrid Models and API Calls

In the internet era, Malware has emerged as one of the significant security threats. Traditional signature-based malware detection techniques struggle to detect new variants of malware thus creating a demand for advanced algorithms to detect and classify malware variants. In malware analysis, API call sequences give insight into malware behavior during run-time execution. Monitoring and analyzing these API sequences is crucial in malware detection. The proposed research work has used a Convolutional Recurrent Neural Network (CRNN) based hybrid model along with data augmentation techniques to provide a more comprehensive feature representation. By capturing both spatial and temporal patterns in API call sequences, the proposed approach offers a comprehensive solution to detect and classify malware-related threats in cyber security. It achieves a higher accuracy of 99.0% on the Virsussign dataset of malware as compared to existing methods, indicating improved performance along with better generalization of the model.