Evolutional concept learning from observations through adaptive feature selection and GA-based feature discovery

This paper presents a method for concept formation of a personal learning a pprentice (PLA) system that attempts to capture users' internal conceptual structure by observing interactions between user and system. The primary go al of a PLA system is to identify the users' cognition that underlies the t aking of action. This is based on the capability to reconstruct internal co ncepts as behavior-shaping constraints by observing operations as well as t he information presented by the system. Our proposed algorithm comprises tw o processes; adaptive feature selection and GA-based feature discovery. The former selects the essential attributes out of a provided set of attribute s that may initially be either relevant or irrelevant, and the latter const ructs new attributes using genetic algorithms applied to a set of elementar y features logically represented in a disjunctive normal form. Our method c an be applied to artificial data as well as to a data set obtained from hum an-machine interactions observed during operation of a simulator of a gener ic dynamic production process.