Brain computation system that searches for internal procedures by functional parts combination

A learning architecture for intelligent system based on brain system nature is proposed. To reduce large amount of learning time that is necessary for conventional methods, we introduce reuse of knowledge that are acquired through past experiences. The architecture enables immediate adaptation to a new task that is similar to well experienced tasks. The effect is evaluated in a series of navigation tasks that becomes difficult stepwise. keywords: Knowledge Reuse, Immediate adaptation, Neural Network, Genetic Algorithm, Navigation