Optimization Techniques for Automated Adaptive Training Systems.

摘要

As adaptive training systems are developed, the big problem encountered is the development of an adaptive logic. Current systems develop their branching schemes without direct usage of learning models. The literature presently contains several new techniques for optimizing instruction. These techniques make use of current learning models by which to make trial-by-trial estimates of the student's progress in learning. This information is then used to solve a set of equations which would select an optimal instructional alternative. Optimal decisions are made by selecting the instructional alternative for which the largest marginal gain in learning is predicted. The present task was to review the techniques available which present the greatest feasibility for applications in the developing training systems. The various optimization techniques selected were presented in their most general form so that the variety of their applications might be apparent. It was concluded that the optimization techniques reviewed were quite feasible and have many powerful options to offer. (Author)