Functional link artificial neural network for multi-label classification

Abstract

In this article, a multi-label functional link artificial neural network (MLFLANN) has been developed to efficiently perform multi-label data classification. The input data is functionally expanded to a higher dimension, followed by iterative learning of the multi-label FLANN (MLFLANN) using the training set. The architecture of the network is less complex and the input space dimension is improved in an attempt to overcome the non-linear nature of the multi-label classification problem. The method has been validated on various multi-label datasets and the results are found to be encouraging.