The neural networks applications in different areas take long and sometimes costly expertise as for example in medicine, etc, given that some features seem to bear no relevance to outputs. This paper considers the possibility of reducing the input space size by discarding irrelevant features in input vectors based on the input sensitivity. To get the computing of the respective sensitivities free of the output determination, the paper introduces normalised input sensitivities in relation with multilayer feedforward (MLP) neural networks.
All articles published by Studies in Informatics and Control (SIC) are published in accordance with the terms and conditions of the Creative Commons Attribution License (CC BY-NC-SA 4.0).