Co-operative learning in cognitive systems

V. Plekhanova

Matematičeskoe modelirovanie, Tome 15 (2003) no. 12, pp. 109-117

Résumé

Contemporary approaches to the formation of co-operative systems ignore a consideration of capability/compatibility factors of their agents. In fact, capabilities/compatibilities define the organisation and performance of co-operative systems. We consider a co-operative system as a cognitive system, which consists of a number of knowledge-interrelated cognitive agents. Interrelation is defined by the agents' available knowledge/skill capabilities with respect to the required capabilities for the learning/performance of tasks. We define co-operative learning as learning from series interactions with the other agents in the cognitive system. Each cognitive agent has to be not only capable of learning its task(s) but also to be compatible with the other agents in the system in order to ensure co-operative learning/performance. We consider knowledge/skill factors as critical variables in co-operative learning. We address problems in the formation of a cognitive system that is capable of co-operative learning. We also identify conditions for co-operative learning in cognitive systems, define knowledge integration as a major task in formation of co-operative cognitive systems and present a formal approach to the engineering of cognitive structure in co-operative cognitive systems.