Approximate Knowledge Extraction using MRA for TYPE-I Fuzzy Neural Networks

Using neural network, we try to model the unknown function/for given input-output data pairs. The connection strength of each neuron is updated through learning. Repeated simulations of crisp neural network produce different values of weight factors that are directly affected by the change of different parameters. We propose the idea that for each neuron in the network, we can obtain reduced model with higher efficiency using wavelet based multi-resolution analysis (MRA) to form wavelet based quasi fuzzy weight sets (WBQFWS) through repeated simulation of the crisp neural network. Such type of WBQFWS provides good initial solution for training type-I fuzzified neural networks. As real data is subjected to noise and uncertainty, therefore, WBQWFS help in the simplification of the structure of the complex problems using low dimensional data sets. Such fuzzy sets are also supportive in approximating the sum of knowledge that a hidden or output neuron contains in the learning frameworks