Efficient fast learning automata

  • Authors:
  • M. S. Obaidat;G. I. Papadimitriou;A. S. Pomportsis

  • Affiliations:
  • Department of Computer Science, Monmouth University, West Long Branch, NJ;Department of Informatics, Aristotle University, Box 888, 54006 Thessaloniki, Greece;Department of Informatics, Aristotle University, Box 888, 54006 Thessaloniki, Greece

  • Venue:
  • Information Sciences—Informatics and Computer Science: An International Journal
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

A new class of learning automata which are capable of supporting high-speed real-time applications is introduced. The proposed learning automata have a unique characteristic: they are capable of performing both probability updating and action selection with a computational complexity which is independent of the number of actions. Apart from their low computational complexity, the proposed automata are capable of achieving a high performance when operating in non-stationary stochastic environments.