Abstract
The paper introduces an exclusion/inclusion fuzzy classification neural network. The network is based on our GFMM [3] and it allows for two distinct types of hyperboxes to be created: inclusion hyperboxes that correspond directly to those considered in GFMM, and exclusion hyperboxes that represent contentious areas of the pattern space. The subtraction of the exclusion hyperboxes from the inclusion hyperboxes, implemented by EFC, provides for a more efficient coverage of complex topologies of data clusters.