Estimation and approximation bounds for gradient-based reinforcement learning

Journal of Computer and System Sciences

Volumn 64, Issue 1, 2002, Pages 133-150

  Bartlett, Peter L ^a   Baxter, Jonathan ^b  

^a BIOwulf Technologies   (United States)

^b WhizBang Labs   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

ALGORITHMS; APPROXIMATION THEORY; ARTIFICIAL INTELLIGENCE; CONVERGENCE OF NUMERICAL METHODS; MARKOV PROCESSES; MATHEMATICAL MODELS; REINFORCEMENT; SET THEORY;

APPROXIMATION BOUNDS;

DECISION THEORY;

EID: 0036477347     PISSN: 00220000     EISSN: None     Source Type: Journal    
DOI: 10.1006/jcss.2001.1793     Document Type: Article

References (20)

1. Stochastic optimization
2. Gradient descent for general reinforcement learning
3. On some algorithms for infinite-horizon policy-gradient estimation
4. Learning dynamical systems in a stationary environment
5. Algorithms for sensitivity analysis of Markov chains through potentials and perturbation realization
6. Stochastic approximation for Monte-Carlo optimization
7. Probability inequalities for sums of bounded random variables
8. Reinforcement learning in POMDPs with function approximation
9. Actor-critic algorithms
10. Nonparametric time series prediction through adaptive model selection
11. Minimum complexity regression estimation with weakly dependent observations
12. Sensitivity analysis via likelihood ratios
13. Policy gradient methods for reinforcement learning with function approximation
14. Simple statistical gradient-following algorithms for connectionist reinforcement learning