Should one compute the Temporal Difference fix point or minimize the Bellman Residual? The unified oblique projection view.
Bruno ScherrerPublished in: ICML (2010)
Keyphrases
- temporal difference
- policy iteration
- policy evaluation
- fixpoint
- reinforcement learning
- td learning
- function approximation
- evaluation function
- model free
- logic programs
- sample path
- least squares
- monte carlo
- action selection
- reinforcement learning algorithms
- markov decision processes
- deductive databases
- step size
- finite state machines
- average reward
- variance reduction
- optimal policy
- function approximators
- state space
- dynamic programming
- finite state
- convergence rate
- markov decision problems
- cost function
- radial basis function
- machine learning