ATMoS+: Generalizable Threat Mitigation in SDN Using Permutation Equivariant and Invariant Deep Reinforcement Learning.
Hauton TsangIman AkbariMohammad A. SalahuddinNoura LimamRaouf BoutabaPublished in: IEEE Commun. Mag. (2021)
Keyphrases
- reinforcement learning
- function approximation
- moment invariants
- multi agent
- markov decision processes
- affine transformation
- countermeasures
- machine learning
- state space
- matrix valued
- reinforcement learning algorithms
- doubly stochastic
- reinforcement learning methods
- model free
- affine invariant
- optimal control
- optimal policy
- supervised learning
- dynamic programming
- risk management
- action selection
- temporal difference
- information security
- privacy issues
- temporal difference learning
- scale space
- invariant properties
- policy search
- sir model