Abstract: The disclosure relates to a method for controlling a driver assistance system during operation of a, especially partially automated, fully automated or autonomous, vehicle, wherein the driver assistance system comprises: a, for example at least one, sensor for observing an environment of the vehicle and an electronic control unit using a, especially at least one, neural network for analyzing sensor data of the sensor and providing perception tasks based on the analyzed sensor data, the method comprising: providing a data set of the sensor data by the sensor) in a spatial domain; transforming the data set of the sensor data by the electronic control unit using frequency analysis into a frequency spectrum in a frequency domain; and analyzing the frequency spectrum of the data set in order to detect an adversarially attacked data set, for example before analyzing the sensor data for providing perception tasks.

Abstract: A method, system and computer program for processing image data, to a vehicle comprising such a system, and to a method, system and computer program for generating a filter. Image data processing may include obtaining the image data, and applying a filter on the image data to generate filtered image data, the filter being configured to suppress adversarial perturbations within the image data. The filtered image data is processed using a machine-learning model.

Abstract: The present invention relates to a method, a computer program, and an apparatus for training a neural network for semantic image segmentation. The invention further relates to an in-car control unit or a backend system, which make use of such a method or apparatus, and to a vehicle comprising such an in-car control unit. In some embodiments and in a first step, image data of a sequence of image frames are received. Then a frame-based evaluation of semantic segmentation predictions of one or more objects in individual image frames is performed. Furthermore, a sequence-based evaluation of temporal characteristics of semantic segmentation predictions of said one or more objects in at least two image frames is performed. The results of the frame-based evaluation and the sequence-based evaluation are combined.

Abstract: The disclosure relates to methods for compressing a neural network, wherein members of a vehicle fleet locally execute the neural network and during at least one inference phase each determine a selection of elements of the neural network that should be pruned, wherein the members of the fleet transmit the respective determined selection to a central server, wherein the central server merges the respective transmitted selections and generates a merged selection, and wherein the central server prunes the neural network on the basis of the merged selection.

Abstract: A method for detecting an adversarial perturbation in input data of a neural network, wherein a conditional generative adversarial network is trained during a training phase, wherein a generator network of the conditional generative adversarial network is trained to generate adversarial perturbations conditioned on input data of the neural network, and wherein a discriminator network of the conditional generative adversarial network is trained at least to detect an adversarial perturbation in the input data generated by the generator network, and wherein, during an application phase, the trained discriminator network detects an adversarial perturbation in input data of the neural network and provides a detection result. Also disclosed is a backend server, a detection device and a system.