“…Approaches based on threat models [46], ontologies and big data techniques [47], static and dynamic analysis [48], Petri nets [49], fuzzy classifiers [50], expert systems [51], planning automation [52], automated pathfinding [53] (including map/graph-based techniques [54]), and agent-based modeling [55] have been proposed. Prior research has also developed a variety of machine learning techniques, such as those using reinforcement learning [56][57][58][59][60] and deep reinforcement learning [61,62], as well as supporting technologies, such as a description language [63]. Despite all of this, some suggest that human penetration testing is still needed due to the capability of human testers to think creatively.…”