Adaptive Secure Threat Intelligence Infrastructure for AI and the Edge

Abstract

Cyber security warfare is actively fought on both sides by AI and machine learning, which allows both adversaries and defenders to engage at unprecedented speeds and scales. For intrusion detection system intelligence, artificial intelligence and machine learning are essential for managing the huge amount of information and assuring the credibility of such data. Dealing with massive amounts of data transitions can lead to major issues termed security challenges. A new learning strategy called Federated Learning (FL) facilitates deep neural network training between numerous dispersed edges nodes without the need for transmitting data thus addressing privacy concerns. Using a federated forest algorithm, a safer cross regional deep learning system is put forward that enables training to be collectively supervised over edge nodes in various regions locations using the same node sample set and yet different feature sets, collecting and analyzing data retained in every one of them while not transferring their original information.