A method for improving paired collaborative learning through approaches of computational intelligence

In school education there are many kinds of learning styles, and it is known that group learning (collaborative learning) is more effective than individual learning. In collaborative learning it is very important how to determine the optimal combination of students in order to improve the learning effect. In this paper we propose a method to improve the learning effect of collaborative learning. A neural network model is first applied for predicting learning results of pairs of students in collaborative learning. Then, in order to determine the optimal pairs of students, a genetic algorithm is applied with the prediction results obtained from the neural network. Based on this combination of students, we carried out an experiment of collaborative learning at a college in Japan. It was confirmed from the experimental results that the proposed method was effective.