Asymptotic behaviour of a learning algorithm
Authors:
- 1 April 1984
- journal article
- research article
- Published by Taylor & Francis in International Journal of Control
- Vol. 39 (4) , 827-838
Abstract
The paper considers a learning automaton operating in a stationary random environment. The automaton has multiple actions and updates its action probability vector according to the linear reward — ϵ penalty (LR-ϵp) algorithm. Using weak convergence concepts it is shown that for large time and small values of parameters in the algorithm, the evolution of the action probability can be represented by Gauss-Markov diffusion.Keywords
This publication has 6 references indexed in Scilit:
- A Learning Model for Routing in Telephone NetworksSIAM Journal on Control and Optimization, 1982
- Averaging Methods for the Asymptotic Analysis of Learning and Adaptive Systems, with Small Adjustment RateSIAM Journal on Control and Optimization, 1981
- Learning Algorithms Theory and Applications1981
- Bounds on the Convergence Probabilities of Learning AutomataIEEE Transactions on Systems, Man, and Cybernetics, 1976
- Learning Automata - A SurveyIEEE Transactions on Systems, Man, and Cybernetics, 1974
- Markovian Learning ProcessesSIAM Review, 1974