Schema-based architectures of machine learning

Schema-based architectures (SBAs) consist of collections of modularly and hierarchically organized schemas, which constitute building blocks for perception, cognition, and action. An SBA organizes these schemas in such a way so that action selection, motor coordination, and cognition in the general sense interact effectively. SBAs were mainly inspired by theories of sensorimotor adaptation and cognitive development and learning. Particularly Jean Piaget's research on and theories of cognitive development in infants and children inspired the design of SBAs. Machine learning develops algorithms to learn, structure, and continuously adapt SBAs. Various forms of representations are used to develop SBAs, including symbolic representations, rule-based representations, as well as neural network representations.