IREA-based Fuzzy Neural Network and Its Application to Network Congestion Prediction
Yang Jing-yu · Acta Simulata Systematica Sinica · 2004
In this paper, rough set theorys ability of extracting crude domain knowledge in the form of rules from the data and FNNs ability of reasoning are combined to improve peoples ability of dealing with uncertainty, imprecision data. An incremental rule extraction algorithm (IREA) is utilized to construct IREA-based FNN. Compared with the classical FNN, IREAbased FNN has characteristics of fewer rules and shorter rule length. Network congestion prediction simulation demonstrates the superiority of the proposed IREA-based FNN over the classical FNN under the circumstance of no initial field knowledge.