Temporal-Difference Networks for Dynamical Systems with Continuous Observations and Actions.
Christopher M. VigoritoPublished in: UAI (2009)
Keyphrases
- dynamical systems
- predictive state representations
- temporal difference
- past observations
- action selection
- partially observable
- reinforcement learning
- td learning
- evaluation function
- action space
- function approximation
- connectionist networks
- state space
- nonlinear dynamical systems
- linear dynamical systems
- monte carlo
- step size
- model free
- stochastic systems
- reinforcement learning algorithms
- policy evaluation
- neural network
- supervised learning
- decision making
- learning algorithm
- function approximators
- reinforcement learning methods
- decision theoretic
- temporal difference methods
- policy iteration
- markov decision processes
- active learning
- objective function
- multiscale
- training data
- machine learning
- data mining