Scalable Online Planning via Reinforcement Learning Fine-Tuning.
Arnaud FickingerHengyuan HuBrandon AmosStuart J. RussellNoam BrownPublished in: NeurIPS (2021)
Keyphrases
- fine tuning
- reinforcement learning
- fine tune
- viable alternative
- online learning
- fine tuned
- heuristic search
- planning problems
- action selection
- real time
- decision theoretic
- reward shaping
- deterministic domains
- macro actions
- partially observable
- function approximation
- complex domains
- learning process
- temporal difference
- blocks world
- decision support
- mobile robot
- multi agent
- learning algorithm