Visual RBF network design based on Star Coordinates

The performance of neural networks is greatly affected by their design, yet the question of finding optimal designs remains open and inspires a considerable amount of research. Most of the researches have been focused on developing automatic algorithms for neural network configuration. This paper addresses the problem of Radial Basis Function Network (RBFN) design definition with a visual technique, called Star Coordinates. The purpose of this approach is to enable the RBFN design revision and refining process, capitalizing on the power of visualization and interactive operations.