How one learns strategies: processes and representation of strategy acquisition
Yuichiro Anzai · 1978
A general theory is presented for the process and representation of strategy acquisition by human and artificial intelligence systems. The process is characterized by inductive generalization and organization of new rules through successive construction of new problem spaces. Various subtasks involved in the process can be naturally represented by subroutines in a hierarchical adaptive production system (HAPS), which is proposed as a representational scheme for strategy acquisition. HAPS has capabilities for subroutine calls and creating new productions in any subroutine. Results from strategy acquisition in the Tower of Hanoi puzzle by a HAPS program are given.