These days machine learning methods and deep learning approaches have revolutionary changed decision making process including the cloud security domain. Deep learning is not a universal tool with the ability to solve all the cloud security problems due to the requirement of large sized training datasets (Najafabadi et al., 2015). Still, there are numerous cloud security problems where the deep learning networks have shown noteworthy enhancements to the traditional security solutions (Papernot et al., 2016). Few cloud security problems are given below where deep learning models have displayed noteworthy enhancements over the traditional machine learning-based approaches (Deng and Yu, 2014; Liu et al., 2017):
Network intrusion detection such as scanning, spoofing etc.;
Phishing attacks (malicious URL identification);
Application attack identification such as OWASP-Top 10 attacks;
Malware identification and categorization;
Ransomware, spyware recognition;
User behaviour study;
Suspicious sign-in activity detection;
Brute force attack detection and other cloud security related problems.
A network intrusion recognition system supports system to detect network security breaches in the business organizations. Hodo et al. (2017) have presented a nontor traffic detection scheme using the support vector machine and artificial neural network based machine learning models. They have also used correlation based feature selection scheme to reduce the feature dimension.