SEED RL: Scalable and Efficient Deep-RL with Accelerated Central Inference.
Lasse EspeholtRaphaël MarinierPiotr StanczykKe WangMarcin MichalskiPublished in: ICLR (2020)
Keyphrases
- reinforcement learning
- function approximation
- highly scalable
- action selection
- lightweight
- neural network
- partially observable domains
- learning agents
- memory efficient
- machine learning
- autonomous learning
- web scale
- complex domains
- markov decision processes
- optimal policy
- state space
- bayesian inference
- learning classifier systems
- decision theoretic
- temporal difference
- computationally efficient
- supervised learning
- efficient learning
- action space
- dynamic programming
- website
- belief nets
- genetic algorithm