Neuro-fuzzy control and modeling in an adaptive information visualization system

In this paper, the neuro-fuzzy technology is considered to model and control an adaptive information visualisation system. Three kinds of neural network used are: a general regression neural network (GRNN), a Kohonen's self-organized feature map (SOFM), and a fuzzy perceptron. The GRNN has been used to constantly predict the time of the redisplay dynamics of an adaptive information visualisation system, the Kohonen SOFM has been used for automatic generation of a set of initial membership functions and an initial fuzzy rule matrix, and the fuzzy perceptron has been used for controlling the redisplay dynamics of the system. The experimental results demonstrate the effectiveness and implementation of the proposed neuro-fuzzy control system. In particular, its performance of information redisplay has been improved up to 13% in terms of quantity of information per unit of time.