Inspired by the paradigm of the backward stepwise regression, we have initially considered that all attributes are potentially important explanatory variables for the class label. Therefore, we have firstly included all of these attributes in the full model (i.e., IC-SLFN), and then removed unimp...