Detection of Malware in Cloud Computing using Sparse Autoencoders

Cloud computing is a most inclining world view that gives conveyance of physical and intelligent assets as administrations over the Internet on request. Numerous malwares focus at customized personal computers (PCs) in cloud condition to obtain secret data and obstacle the cloud appropriation by organizations and clients. In this paper, we consider a way to deal with shielding the cloud from being assaulted by nearby PCs. Because of this issue, in view of the Windows Application Programming Interface (API) calls are removed from the Portable Executable (PE) files, we propose a novel Behavior-based Machine Learning Framework (BMLF) using Sparse Autoencoder (SpAE) which is worked in cloud stage for detection of malware. In the proposed BMLF, first we develop conduct graphs to give effective data of malware practices utilizing extricated through. We at that point utilize SpAEs for removing elevated level highlights from conduct graphs. The layers of SpAEs are embedded in a steady progression and the last layer is associated with an additional classifier. The design of SpAEs is 5,000-2,000-1000. The experimental results show that the proposed BMLF yields the semantics of more elevated level noxious practices and increments the normal detection accuracy by 2%.