US 11,699,075 B2
Method for training an artificial neural network, artificial neural network, use of an artificial neural network, and corresponding computer program, machine-readable memory medium, and corresponding apparatus
Oliver Willers, Korb (DE); and Sebastian Sudholt, Neuenstadt Am Kocher (DE)
Assigned to ROBERT BOSCH GMBH, Stuttgart (DE)
Filed by Robert Bosch GmbH, Stuttgart (DE)
Filed on Jun. 25, 2020, as Appl. No. 16/911,681.
Claims priority of application No. 10 2019 209 457.0 (DE), filed on Jun. 28, 2019.
Prior Publication US 2020/0410342 A1, Dec. 31, 2020
Int. Cl. G06N 3/08 (2023.01); G06N 20/20 (2019.01); G06N 7/01 (2023.01); G06V 10/764 (2022.01); G06V 20/56 (2022.01)
CPC G06N 3/08 (2013.01) [G06N 7/01 (2023.01); G06N 20/20 (2019.01); G06V 10/764 (2022.01); G06V 20/56 (2022.01)] 13 Claims
OG exemplary drawing
 
12. A non-transitory machine-readable memory medium on which is stored a computer program, the computer program, when executed by a computer, causing the computer to perform the following steps:
providing an artificial neural network for controlling a technical system, the artificial neural network being trained using training data sets, including adaption of a parameter of the artificial neural network depending on a loss function, the loss function including a first term that represents an estimate of a lower bound of distances between classifications of the training data sets by the artificial neural network and expected classifications of the training data sets, wherein the loss function includes a second term that is configured in such a way that differences in aleatoric uncertainty in the training data sets over different samples of the artificial neural network are regulated; and
controlling the technical system using the trained artificial neural network, the technical system including a robot, or a vehicle, or a tool, or a machine.