The fastest Learning Automata (LA) algorithms currently available fall in the family of Estimator Algorithms introduced by Thathachar and Sastry. The pioneering work of these authors was the Pursuit Algorithm, which pursues only the current estimated optimal action. Later, the same authors introduced a more sophisticated estimator algorithm, known as the TSE algorithm. This paper introduces first a vectorial representation the TSE algorithm that shows more clearly the underlying concepts of the TSE algorithm. Furthermore, using this vectorial representation, we introduce a Generalized TSE estimator algorithm (GTSE). We argue that this learning scheme minimizes the probability of pursuing a wrong action and it is proven empirically to be the fastest converging estimator learning algorithm known to date. To attest this, we present a quantitative comparison of its performance against the TSE and other existing continuous estimator algorithms.

Additional Metadata
Keywords Estimator Algorithms, Learnng Automata, Pursuit Algorithms, TSE
Conference 2004 IEEE Conference on Cybernetics and Intelligent Systems
Citation
Agache, M. (Mariana), & Oommen, J. (2004). Generalized TSE: A new generalized estimator-based Learning Automaton. In 2004 IEEE Conference on Cybernetics and Intelligent Systems (pp. 245–251).