Abstract: Security issues in the Internet of Things (IoT) ecosystems are becoming a major concern for users, developers and business owners. The traditional security models are not full proof enough to handle security issues in a ubiquitous environment. Intelligent mechanisms have been developed to address loopholes in the security of network systems; nonetheless, cybercrimes have increased in the computing ecosystems due to increased surface of attacks created by the adoption of IoT. In the recent years Fog computing has been adopted to decentralize application and service provision. In this chapter we shed light on Fog computing security architecture. We concentrate on the role of machine learning (ML) in mitigating issues of security. We present a study that underlines the next-generation secure fog infrastructure. We further prompt concerns about threats, vulnerabilities and exploits in fog-cloud of things. In our work, we cautiously look at ML-based botnet detection, authentication, access control, botnet detection, malware detection and classification, and offloading. Lastly, this chapter discusses applications, opportunities, challenges and future trends.