Learning of hierarchical fuzzy aggregative network using simplified swarm optimization

Hierarchical fuzzy aggregation network (HFAN) is a fine multilayer information fusion system that carries out multi-criteria aggregation. It can be regarded as a functional classifier for dealing with decision-making problems. The HFAN comprises fuzzy aggregation operators built from adjusted parameters (γ) and associated weights (δ). In this paper, we adopt soft computing techniques (e.g., PSO and SSO) to learn these fuzzy aggregation operators. We provide association rules to define input data and use a hierarchical clustering algorithm to organize the network structure. The optimization efficiency of these rules is experimented with different network topologies and datasets. We verify effectiveness of HFAN by applying it to classify the breast cancer dataset from the UCI Machine Learning Repository. We conduct study for comparing the optimized HFAN and other approaches in terms of ten-fold cross-validation.