Reinforcement Learning for feedback-enabled cyber resilience

Leveraging Computational Intelligence Techniques for Defensive Deception: A Review, Recent Advances, Open Problems and Future Directions

“…DNN (Deep Neural Networks) consists of neural networks with a big number of disguised layers [ 15 ]. DNNs are a subset of ANNs.…”

“…Deep Reinforcement Learning (DRL) estimates difficult functions with high-dimensional inputs using a neural network. The addition of deep learning to traditional RL approaches improves the ability to capture the huge scale of numerous Internet-connected systems, such as mobile networks and IoT devices [ 15 ]. DRL could develop a low-dimensional version of high-dimensional data, which is quite compact in nature.…”

“…DRL could develop a low-dimensional version of high-dimensional data, which is quite compact in nature. DNN is a powerful complement, allowing it to be used for the cyber security of vast networked systems [ 15 ]. There has been a recent surge of research on using RL to select an adaptive configuration strategy to maximize the impact of MTD, with particular emphasis on the dynamic environment, reduced resource consumption, usability, partially observable environments, and multi-agent scenarios that include both the system’s characteristics and the adversary’s observed activities.…”

“…There has been a recent surge of research on using RL to select an adaptive configuration strategy to maximize the impact of MTD, with particular emphasis on the dynamic environment, reduced resource consumption, usability, partially observable environments, and multi-agent scenarios that include both the system’s characteristics and the adversary’s observed activities. Through a compromise between usability and security, we can investigate the potential of RL in reconfiguring defenses [ 15 ]. Creating hybrid Defensive Deception tactics that incorporate machine learning and game theory: Other protection mechanisms have been considered using machine learning-based game-theoretic techniques.…”

“…Both insider and outsider attacks can be prevented using Deception. These days machine learning has emerged as an effective technology that provides us with a wide range of applications ranging from recognition of patterns, image identification, image, and video processing, making predictions, virus or malware detection, autonomous driving, and other application scenarios [ 8 , 9 , 10 , 11 , 12 , 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 , 30 , 31 , 32 , 33 , 34 , 35 , 36 , 37 , 38 , 39 ]. The advantages of machine learning algorithms can be extended for deploying Defensive Deception frameworks [ 40 , 41 , 42 , 43 , 44 , 45 , 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 , 54 , 55 , 56 , 57 , 58 , …”