GNN-RL: Dynamic Reward Mechanism for Connected Vehicle Security using Graph Neural Networks and Reinforcement Learning.
Heena RathoreHenry GriffithPublished in: SMARTCOMP (2023)
Keyphrases
- reinforcement learning
- neural network
- function approximation
- state space
- function approximators
- model free
- reinforcement learning algorithms
- dynamic environments
- learning capabilities
- temporal difference
- connected components
- machine learning
- learning classifier systems
- dynamic programming
- back propagation
- eligibility traces
- control policy
- learning algorithm
- multi agent
- artificial neural networks
- access control
- learning problems
- optimal policy
- direct policy search
- information security
- partially observable
- complex domains
- rl algorithms
- markov decision process
- connected graphs
- partially observable environments
- dynamic graph
- intrusion detection
- random walk
- learning agent
- optimal control
- directed graph
- reward function
- learning process
- action selection
- sufficient conditions