Temporal-Difference Networks for Dynamical Systems with Continuous Observations and Actions
Christopher M. VigoritoPublished in: CoRR (2012)
Keyphrases
- dynamical systems
- predictive state representations
- temporal difference
- past observations
- action selection
- reinforcement learning
- partially observable
- action space
- state space
- connectionist networks
- td learning
- function approximation
- evaluation function
- step size
- reinforcement learning algorithms
- model free
- nonlinear dynamical systems
- policy evaluation
- monte carlo
- partially observable markov decision processes
- stochastic systems
- linear dynamical systems
- policy iteration
- real valued
- supervised learning
- temporal difference methods