Applying Neural Networks Approach to Achieve the Parameter Optimization for Censored Data

This study proposes an approach based on neural networks to perform the analysis of the censored data. Two neural networks are constructed: the first neural network is designed to estimate the censored data by means of constructing the model derived from the uncensored data and, the second neural network is designed to obtain the optimum parameter settings of control factors by using the uncensored data and the estimated censored data. The proposed approach can not only be used for control factors with categorical level setting, but also for continuous setting. Furthermore, the approach proposed herein will be employed for Taguchi’s dynamic experiment. One numerical example demonstrates the effectiveness of the proposed approach. Key-Words: censored data, neural network, parameter optimization