Training for artificial neural networks with better utilization of the learning data sets

摘要

Method (100) for training an artificial neural network, ANN (1), with learning data sets (2) each comprising a sample (11a) and an associated label (12a), each sample (11a) having values of one or more Input variables (11) of the ANN (1) and the associated label (12a) represents values of one or more output variables (12) of the ANN (1), with the following steps: • The ANN (1) is based on the in the learning data records (2) contained samples (11a) are trained (110) unsupervised; • using at least one quality criterion (3) for the reliability and / or quality of the assignment of labels (12a) to samples (11a) in the learning data sets (2) a weighting (2a) of the learning data records (2) is determined (120); • taking this weighting (2a) into account, the ANN (1) is calculated using the samples (11a) and associated labels (11a) contained in the learning data records (2). 12a) monitors trained (130). Trained ANN (1 *) and system (50) with this ANN (1 *). Associated computer p rogram.