Adaptive and Learning Systems

A DAPTIVE and learning systems have drawn considerable attention in the last few decades due to their inherent strength in suitably modeling many real-world complex systems , which are otherwise difficult to model using the traditional existing tools and systems. There exist many theories of adaptation and learning such as learning automata (LA), neural networks, and cellular automata. Considering the case of LA, for example, although it is only quite recently that LA and their applications in solving complex problems have become popular, their history dates back to the 1950s and 1960s with reference to the works of mathematicians and mathematical psychologists such as Bush and Mosteller, Atkinson et al., Tsetlin, and Varshavskii and Vorontsova, among others. Some of the popular current-generation researchers on LA include In typical LA systems, a self-operating machine or a mechanism , termed as an Automaton, responds to a sequence of instructions in a certain way, so as to achieve a certain goal. The Automaton either responds to a predetermined set of rules or adapts to the environmental dynamics in which it operates. The term learning has its root in Psychology and is used to refer to the act of acquiring knowledge and modifying one's behavior based on the experience gained. Thus, in LA, the adaptive automaton adapts to the responses from the environment through a series of interactions with it. It then attempts to learn the best action from a set of possible actions that are offered to it by the random stationary or nonstationary environment in which it operates. The Automaton thus acts as a decision-maker to arrive at the best action. Some of the attractive features of LA such as their ability to rapidly and accurately converge and their low computational complexity have made them useful for solving problems involving network call-admission control, distributed scheduling, training hidden Markov models, neural-network adaptation, graph partitioning, intelligent vehicle control, dynamic shortest path, and pattern classification. Their advantages appear prominent in optimizing problems in which an optimal action needs to be determined from a set of actions. Typically, learning is of best help only when there are high levels of uncertainty in the system in which the automaton operates. This Special Issue of IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS—PART B has attracted many papers from researchers all over the world who are currently active in research in stochastic learning systems, so that their research results …