Regularized training of neural networks

摘要

Method (100) for training an artificial neural network, ANN (1), which translates one or more input variables (11) into one or more output variables (13) by means of learning data sets (2), the learning input variable values (11a) Measurement data and associated learning output variable values (13a) include, with the following steps: learning input variable values (11a) from at least one learning data set (2) are mapped (110) by the ANN (1) onto output variable values (13); the output variable values (13) from the respective learning output variable values (13a) are processed (120) in accordance with a cost function (14) to a measure for the error (14a) of the ANN (1) in the processing of the learning input variable values (11a) • The error (14a) becomes changes in the parameters (12) through backpropagation, their implementation during the further processing of learning input variable values (11a) by the ANN (1) the evaluation of the output variable values (13) obtained thereby by the cost radio tion (14) is expected to be improved, determined (130) and applied to the ANN (1) (140); ).