A Comparison of Some Error Estimates for Neural Network Models 论文

1996Neural Computation引用 294
Neural Networks and ApplicationsControl Systems and IdentificationFault Detection and Control Systems

详细信息

发表期刊/会议
Neural Computation
发表日期
1996-01-01
发表年份
1996

关键词

Neural Networks and ApplicationsControl Systems and IdentificationFault Detection and Control Systems

摘要

We discuss a number of methods for estimating the standard error of predicted values from a multilayer perceptron. These methods include the delta method based on the Hessian, bootstrap estimators, and the “sandwich” estimator. The methods are described and compared in a number of examples. We find that the bootstrap methods perform best, partly because they capture variability due to the choice of starting weights.