Using Ternary Rewards to Reason over Knowledge Graphs with Deep Reinforcement Learning.
Fréderic GodinAnjishnu KumarArpit MittalPublished in: CoRR (2019)
Keyphrases
- reinforcement learning
- markov decision processes
- learned knowledge
- function approximation
- domain knowledge
- domain experts
- knowledge sources
- partially observable
- knowledge representation
- state space
- supervised learning
- temporal difference
- knowledge acquisition
- learning systems
- knowledge sharing
- reinforcement learning methods
- graph matching
- data mining techniques
- knowledge management
- dynamic programming
- prior knowledge
- expert systems