Multi-instance Multi-label Text Categorization Algorithm Based on Multi-quadric Function Radial Basis Network Model

The activation function of the hidden layer in MIMLRBF algorithm is a radial basis function which is always a Gaussian function, but the performance of algorithm depends on the center, the width of radical basis function and weight values. In this paper, the radial basis is set as multi-Quadric function which is widely used in interpolation. The parameters in experimental design and multi-Quadric function are determined by Hamming Loss, Ranking Loss, One Error and Coverage. The experimental results in text classification show that the method is highly competitive to other MIML algorithms.