Grid-to-Graph: Flexible Spatial Relational Inductive Biases for Reinforcement Learning.
Zhengyao JiangPasquale MinerviniMinqi JiangTim RocktäschelPublished in: AAMAS (2021)
Keyphrases
- reinforcement learning
- machine learning
- spatial information
- relational reinforcement learning
- relational learning
- relational data
- graph structure
- grid cells
- spatio temporal
- spatial data
- inductive logic programming
- function approximation
- graph model
- weighted graph
- graph kernels
- random walk
- graph theory
- relational databases
- relational structures
- learning algorithm
- graph representation
- directed graph
- space time
- spatial relationships
- data model
- state space
- optimal policy
- undirected graph
- spatial distribution
- graph mining
- directed acyclic graph
- spatial and temporal
- connected components
- inductive learning
- spatial relations
- structured data
- grid computing
- graph matching
- information extraction
- neighborhood graph
- rewriting rules
- knowledge representation