A study on evaluating and improving the reliability of bank note neuro-classifiers

This paper addresses the reliability of the bank note classifiers and a new method is proposed for improving the classification reliability based on the local principal components analysis (PCA). The reliability is evaluated by using an algorithm, which employs a function of winning class probability and second maximal probability in the LVQ classifier. The experimental results from 3,600 data samples show an increase up to 100% in the reliability of classification.