Asymptotic analysis of stochastic approximation algorithms under violated Kushner-Clark conditions with applications

Proceedings of the IEEE Conference on Decision and Control

Volumn 3, Issue , 2000, Pages 2875-2880

  Tadic V ^a  

^a UNIVERSITY OF MELBOURNE   (Australia)

Author keywords

[No Author keywords available]

Indexed keywords

APPROXIMATION THEORY; ASYMPTOTIC STABILITY; FUNCTIONS; LYAPUNOV METHODS; MARKOV PROCESSES; RANDOM PROCESSES; THEOREM PROVING; WHITE NOISE;

ASYMPTOTIC ANALYSIS; KUSHNER-CLARK CONDITIONS; LYAPUNOV FUNCTION; NONLINEAR FUNCTION APPROXIMATION; STOCHASTIC APPROXIMATION ALGORITHMS; TEMPORAL-DIFFERENCE LEARNING ALGORITHMS;

LEARNING ALGORITHMS;

EID: 0034439674     PISSN: 01912216     EISSN: None     Source Type: Conference Proceeding    
DOI: None     Document Type: Conference Paper

References (16)

1. A dynamical system approach to stochastic approximation
2. Convergence and robustness of the Robins-Monro algorithm truncated at randomly varying bounds
3. Robustness analysis for stochastic approximation algorithms
4. Recent developments in stochastic approximation
5. General results on the convergence of stochastic approximation
6. Convergence of stochastic approximation algorithms: From the Kushner-Clark theorem to the Lyapunov functional method
7. Applications of a Kushner and Clark lemma to general classes of stochastic algorithms
8. Asymptotic analysis of stochastic iterative algorithms under violated Kushner-Clark conditions with applications to machine learning
9. An analysis of temporal-difference learning with function approximation
10. Equivalent and sufficient conditions on noise sequences for stochastic approximation algorithms