Grid-to-Graph: Flexible Spatial Relational Inductive Biases for Reinforcement Learning.
Zhengyao JiangPasquale MinerviniMinqi JiangTim RocktäschelPublished in: CoRR (2021)
Keyphrases
- reinforcement learning
- grid cells
- relational data
- spatial information
- machine learning
- spatial and temporal
- spatial data
- random walk
- relational reinforcement learning
- spatio temporal
- data model
- function approximation
- graph structure
- graph matching
- structured data
- state space
- graph model
- directed acyclic graph
- spatial databases
- bipartite graph
- grid environment
- graph theoretic
- relational learning
- graph representation
- spanning tree
- neighborhood graph
- grid points
- graph theory
- inductive logic programming
- directed graph
- inductive learning
- multi agent
- relational databases
- database systems
- database
- dynamic programming
- graphical models
- inductive bias
- relational structures
- peer to peer
- graph construction
- graph kernels
- reinforcement learning algorithms
- connected components
- inductive inference
- graph databases
- grid computing
- learning algorithm
- concept learning
- spatial relations
- spatial relationships