Task-Agnostic Dynamics Priors for Deep Reinforcement Learning.
Yilun DuKarthik NarasimhanPublished in: CoRR (2019)
Keyphrases
- reinforcement learning
- belief nets
- function approximation
- dynamic model
- model free
- robotic control
- reinforcement learning algorithms
- state space
- reinforcement learning methods
- dynamic programming
- multi agent reinforcement learning
- temporal difference learning
- optimal policy
- dynamical systems
- genetic algorithm
- machine learning
- optimal control
- dynamical models
- data sets
- markov decision processes
- multi agent
- neural network
- bayesian framework
- learning problems
- initial conditions
- robot control
- sufficient conditions
- prior knowledge
- highly nonlinear
- data mining
- real time