Mixtures of Experts Unlock Parameter Scaling for Deep RL.
Johan S. Obando-CeronGhada SokarTimon WilliClare LyleJesse FarebrotherJakob N. FoersterGintare Karolina DziugaiteDoina PrecupPablo Samuel CastroPublished in: CoRR (2024)
Keyphrases
- reinforcement learning
- model free
- human experts
- parameter space
- mixture model
- function approximation
- expert finding
- multi agent
- state space
- parameter values
- mixtures of gaussians
- blind source separation
- input parameters
- expectation maximization
- learning algorithm
- parameter settings
- markov decision processes
- gaussian mixture
- dynamic programming
- reinforcement learning algorithms
- deep learning
- genetic algorithm
- neural network