On Simulating Tutorial-Like Systems Using A Learning Automata Philosophy

摘要

The aim of this keynote speech paper is to present a new philosophyrnto simulate Tutorial-like systems using Learning Automatarn(LA). The system simulates a Socratic model of learningrnfor teaching uncertain course material by simulating thernlearning process for both the Teacher and a School of Students.rnFirst of all, the Teacher has to learn what teachingrnmaterial has to be presented. On the other hand, each Studentrnin the School has to learn and acquire the knowledge of thernpresented course material. We propose that the Tutorial-likernsystem be composed of an ensemble of LA models, whererneach entity can improve its behavior based on the response itrnreceives from its corresponding Environment. Thus, the systemrnalso uses LA to model and simulate the learning achievedrnby the Teacher, the learning achieved by each Student by interactingrnwith the Teacher, and the learning accomplished byrnthe School of Students by interacting between themselves. Inrnparticular, the latter interaction permits Students to enhancerntheir learning by sharing knowledge with each other. Suchrna transfer of information (between the Students) is modeledrnby permitting the exchange of the partial or the entire actionrnprobability vector of the corresponding LA. The keynoternspeech will describe the model of the entire system, whichrncould be used to teach real-life or artificial entities. Thus,rnfor example, the system could be used as a Tutor for anotherrnsoftware component. Some initial results from simulating thernsystem confirm that LA can be a powerful mechanism forrnsimulating the different models of a Tutorial-like system.