Learning by Knowledge Sharing in Autonomous Intelligent Systems

Abstract

Abstract. Very few learning systems applied to problem solving have focused on learning operator definitions from the interaction with a completely unknown environment. In order to achieve better learning convergence, several agents that learn separately are allowed to interchange each learned set of planning operators. Learning is achieved by establishing plans, executing those plans in the environment, analyzing the results of the execution, and combining new evidence with prior evidence. Operators are generated incrementally by combining rote learning, induction, and a variant of reinforcement learning. The results show how allowing the communication among individual learning (and planning) agents provides a much better percentage of successful plans, plus an improved convergence rate than the individual agents alone.

Investigación en Progreso: Método de Evaluación Dinámica de Planes en Sistemas Inteligentes Autónomos

“…En los últimos años, se han elaborado varias modificaciones al modelo con el fin de mejorar el rendimiento de su aprendizaje y de su planificación. En [10] se incorporó una arquitectura multiagente que permite el intercambio de teorías; en [15] se utilizó un sistema multiagente pero en este caso se definieron distintos perfiles de agentes, cada uno de los cuales con un determinado método de adquisición y transmisión de conocimiento; en [17] se propuso un mecanismo de ponderación de planes distinto; por último, en [24] se tomó como marco de trabajo el algoritmo de ponderación elaborado por López pero se propuso la aplicación de algoritmos genéticos para mejorar el rendimiento.…”

“…Esta arquitectura de sistema inteligente autónomo también percibe el entorno a través del sistema sensor, pero antes de realizar cualquier acción, se pregunta si es necesario intercambiar operadores con otro sistema inteligente autónomo [10]. Este proceso se lleva a cabo mediante un módulo de intercambio de operadores.…”

Application of Fuzzy Logic in Learning Autonomous Robots Systems

“…This paper extends previous work learning in Intelligent Autonomous Systems [1,2] by implementing fuzzy methods in the comparison of situations (registered by the proximity sensors of the robot) and actions (made by robot actuators or wheels) between theories that form ARS basis of knowledge [3,4,5]. The Learning Life Cycle in Autonomous Robots Systems [1,2] can be considered as a number of perceptions of the surrounding world and actions performed as a response to those perceptions.…”

A cooperative learning method based on cellular learning automata and its application in optimization problems

“…Cooperative learning techniques [26] are a class of techniques that are designed to address coopera- 40 tion between agents in multi-agent environments with the purpose of improving the quality, the accuracy, and the speed of the learning process. Knowledge sharing [27] is one of the cooperative learning techniques in which agents try to learn from each other and share what they learn. We use this technique in our method by letting agents use neighboring agents' knowledge gathered through 45 their individual learning processes.…”