Predictive representations can link model-based reinforcement learning to model-free mechanisms.
Evan M. RussekIda MomennejadMatthew BotvinickSamuel J. GershmanNathaniel D. DawPublished in: PLoS Comput. Biol. (2017)
Keyphrases
- model free
- model based reinforcement learning
- reinforcement learning
- markov decision processes
- function approximation
- reinforcement learning algorithms
- temporal difference
- policy iteration
- dynamic programming
- multi agent
- markov decision problems
- neural network
- optimal policy
- learning algorithm
- partially observable
- action space
- function approximators