A New Multi-attribute Decision Making Method Based on Fuzzy Neural Network

Application of neural networks in fuzzy multi-attribute decision making is studied. A new fuzzy RBF neural network model, which uses triangular fuzzy numbers as inputs, is set up. Another important feature of the model is that the weights of the inputs fully reflect the effect of the decision-maker's preferences for uncertainty on decision results. Further, attribute weights determined by neural networks have the advantages of both subjective and objective weights. Due to the fact that ideal solution samples are introduced into the training samples, the decision results tends to be more in agreement with the decision-maker's intensions and, therefore, more scientific. Numerical illustrations prove the new model to be effective.