NN classifiers: reducing the computational cost of cross-validation by active pattern selection

Abstract

We propose a new approach for leave-one-out cross-validation of neural network classifiers called "cross-validation with active pattern selection" (CV/APS). In CV/APS, the contribution of the training patterns to backpropagation learning is estimated and this information is used for active selection of CV patterns. On two artificial examples, the computational cost of CV can be reduced to 25% of the normal costs with only small or no errors.