by V. S. K. ReddyI. NagarajuMantri GayatriR. RDileep PRevathy P
In the progress of digital era, there is a significant rise in the number of security flaws that are the result of assaults by malicious software, sometimes known as malware. Malware detection is still a popular area of study owing to the fact that a significant number of computer users, organizations, and governments have been impacted by the exponential development in malicious software assaults. These detection methods rely on malware signatures and behavioral patterns. Recent malwares make use of metamorphic, polymorphic, and other evasive strategies to rapidly modify their behavior and produce a huge number of new malwares. These tactics allow the malware to avoid detection. Recent years have seen an increase in the use of Machine Learning Algorithms (MLA) for the purpose of effective malware analysis. The majority of these newly discovered malwares are variations of previously discovered malware. However, the conventional models failed to classify the different types of malwares and resulted in poor classification performance. So, this work focused on implementation of Malware Detection and Classification (MDC-Net). Initially, dataset preprocessing is performed to remove the missing symbols, unknown data from MALIMG dataset. Then, convolutional neural network (CNN) model is used for extraction malware class dependent, and malware class specific features. Finally, Extreme Learning Machine (ELM) model trained with the CNN features. So, the ELM model can capable of classifying the various malware classes from every new data. The MDC-Net resulted in superior performance than existing approaches in terms various metrics.