Precision Teaching Based on Data-driven Model

Abstract

This paper proposed an innovative data-driven approach DNN-based Precision Teaching Model (DNN-PTM) combining teaching strategies, teaching quality and learning effect with deep neural network techniques. We implement Deep Neural Network (DNN) to evaluate learning effect by analyzing teaching data. DNN-PTM aims to provide personalized and adaptive teaching with the characteristics of "precise teaching and student-centered learning". It focuses on developing the dynamic auto-tuning instructions to cater to learning preferences for each student not for the class. Moreover, DNN-PTM can establish a Personal Knowledge Map through three steps: (I) organizing data: to collect massive of explicit data (directly gathered in the process of teaching and learning) and implicit data (indirectly describes the quality of teaching and learning); (II) building model: to analyze the relationship among teaching behaviors, learning characteristics and education results; (III) Evaluating quality: to measure the quality of an optimal PT strategy predicted in (II) according to its positive effects on teaching and learning. Therefore, DNN-PTM has strong adaptability and intelligence because it can learn a best possible teaching decision which is suitable for the current learning situation from a large number of data.