Symbolic Relational Deep Reinforcement Learning based on Graph Neural Networks.
Jaromír JanischTomás PevnýViliam LisýPublished in: CoRR (2020)
Keyphrases
- neural network
- reinforcement learning
- neural learning
- connectionist systems
- graph theory
- connectionist models
- relational reinforcement learning
- pattern recognition
- random walk
- directed graph
- function approximators
- function approximation
- graph structure
- genetic algorithm
- bipartite graph
- graph representation
- relational data
- recurrent neural networks
- neural nets
- symbolic knowledge
- graph model
- data model
- reinforcement learning algorithms
- relational learning
- multi relational
- model free
- graph theoretic
- learning classifier systems
- relational databases
- learning algorithm
- machine learning
- symbolic representation
- fault diagnosis
- spanning tree
- fuzzy logic
- connected components
- supervised learning
- learning process
- relational structures
- neural network model
- learning tasks
- rewriting rules
- structured data
- markov decision processes