FF + FPG: Guiding a Policy-Gradient Planner.
Olivier BuffetDouglas AberdeenPublished in: ICAPS (2007)
Keyphrases
- policy gradient
- heuristic search
- planning domains
- initial state
- planning systems
- actor critic
- single agent
- reinforcement learning
- parametric optimization
- gradient method
- function approximation
- reinforcement learning algorithms
- partially observable markov decision processes
- optimal control
- ai planning
- planning problems
- state space
- orders of magnitude
- model free reinforcement learning
- domain independent
- approximation methods
- search algorithm
- optimal policy
- variance reduction
- state action
- reinforcement learning methods
- situation calculus
- optimization methods
- model checking
- dynamic programming
- cost function
- search space
- multi agent
- neural network