Abstract: A device and a method of training a generative neural network. The method includes: generating an edge image using an edge detection applied to a digital image, the edge image comprising a plurality of edge pixels determined as representing edges of one or more digital objects in the digital image; selecting edge-pixels from the plurality of edge pixels; providing a segmentation image using the digital image, the segmentation image comprising a plurality of first pixels, the positions of the first pixels corresponding to the positions of the selected edge-pixels; selecting one or more second pixels for each first pixel in the segmentation image; generating a distorted segmentation image using a two-dimensional distortion applied to the segmentation image; and training the generative neural network using the distorted segmentation image as input image to estimate the digital image.

Abstract: A system and method are provided for classifying objects in spatial data using a machine learned model, as well as a system and method for training the machine learned model. The machine learned model may comprise a content sensitive classifier, a location sensitive classifier and at least one outlier detector. Both classifiers may jointly distinguish between objects in spatial data being in-distribution or marginal-out-of-distribution. The outlier detection part may be trained on inlier examples from the training data, while the presence of actual outliers in the input data of the machine learnable model may be mimicked in the feature space of the machine learnable model during training. The combination of these parts may provide a more robust classification of objects in spatial data with respect to outliers, without having to increase the size of the training data.

Abstract: A computer-implemented method of training a machine learnable function, such as an image classifier or image feature extractor. When applying such machine learnable functions in autonomous driving and similar application areas, generalizability may be important. To improve generalizability, the machine learnable function is rewarded for responding predictably at a layer of the machine learnable function to a set of differences between input observations. This is done by means of a regularization objective included in the objective function used to train the machine learnable function. The regularization objective rewards a mutual statistical dependence between representations of input observations at the given layer, given a difference label indicating a difference between the input observations.

Abstract: A method for training a neural network. The neural network comprises a first layer which includes a plurality of filters to provide a first layer output comprising a plurality of feature maps. Training of the classifier includes: receiving, by a preceding layer, a first layer input in the first layer, wherein the first layer input is based on the input signal; determining the first layer output based on the first layer input and a plurality of parameters of the first layer; determining a first layer loss value based on the first layer output, wherein the first layer loss value characterizes a degree of dependency between the feature maps, the first layer loss value being obtained in an unsupervised fashion; and training the neural network. The training includes an adaption of the parameters of the first layer, the adaption being based on the first layer loss value.

Abstract: A training method for training a generator neural network configured to generate synthesized sensor data. A fidelity destroying transformation is defined configured to transform a measured sensor data to obtain a fidelity-destroyed transformed measured sensor data. A fidelity preserving transformation is defined configured to transform a measured sensor data to obtain a fidelity-preserved transformed measured sensor data.

Abstract: A training method for training a generator neural network configured to generate synthesized sensor data. A discriminator network is configured to receive discriminator input data comprising synthesized sensor data and/or measured sensor data, and to produce as output localized distinguishing information, the localized distinguishing information indicating for a plurality of sub-sets of the discriminator input data if the sub-set corresponds to measured sensor data or to synthesized sensor data.

Abstract: Device and method of classifying data based on a model which includes a generator and a classifier. The method includes providing a training pair including a training example belonging to a dataset and a class embedding belonging to a first set of class embeddings, training the generator to generate artificial training examples in a feature space depending on the training pair, determining an artificial training example in the feature space depending on the generator and depending on a class embedding of a second set of class embeddings, training the classifier to determine a class for the artificial sample from a set of classes depending on the artificial sample and the class embedding, the set of classes being the union of a first and second set of classes, characterized by the first and second set of class embeddings, respectively, and classifying data depending on the classifier.

Abstract: A computer-implemented method for training an augmented discriminator and a generator. The method includes: providing a training set comprising real training samples and artificial training samples for training of the augmented discriminator, wherein the artificial training samples are generated by the generator; assigning a data sequence to at least one data sample of the training set; wherein each pair of data sample and assigned data sequence is assigned to one of a plurality of classes such that, given the assigned one class and the assigned data sequence characterize whether the data sample is a real training sample or an artificial training sample; training the augmented discriminator to compute from pairs of data sample and assigned data sequence the respective class to which the corresponding pair is assigned; training the generator to generate artificial training samples such that the augmented discriminator is not able to correctly compute the aforementioned one class.

Abstract: A method for generating synthetic measurement data indistinguishable from actual measurement data captured by a first physical measurement modality. The first physical measurement modality is based on emitting an interrogating wave towards an object and recording a reflected wave coming from the object in a manner that allows for a determination of the time-of-flight between the emission of the interrogating beam and the arrival of the reflected wave. The method includes: obtaining a first compressed representation of the synthetic measurement data in a first latent space, wherein this first latent space is associated with a first decoder that is trained to map each element of the first latent space to a record of synthetic measurement data that is indistinguishable from records of actual measurement data of the first physical measurement modality, and applying the first decoder to the first compressed representation, so as to obtain the sought synthetic measurement data.

Abstract: Device and method for training an artificial neural network, including providing a neural network layer for an equivariant feature mapping having a plurality of output channels, grouping channels of the output channels into a number of distinct groups, wherein the output channels of each individual distinct group are organized into an individual grid defining a spatial location of each of the output channels of the individual distinct group in the grid for the individual distinct group, providing for each of the output channels of each individual distinct group, a distinct normalization function which is defined depending on the spatial location of the output channel in the grid in that this output channel is organized and depending on tunable hyperparameters for the normalization function, determining an output of the artificial neural network depending on a result of each of the distinct normalization functions, training the hyperparameters of the artificial neural network.

Abstract: A generator for generating three-dimensional point clouds of synthetic LIDAR signals from a set of LIDAR signals measured with the aid of a physical LIDAR sensor. The generator includes a random generator and a first machine learning system, which receives vectors or tensors of random values from the random generator as input, and maps each such vector, or each such tensor, onto a three-dimensional point cloud of a synthetic LIDAR signal with the aid of an internal processing chain. The internal processing chain of the first machine learning system is parameterized by a plurality of parameters which are set in such a way that the three-dimensional point cloud of the LIDAR signal, and/or at least one characteristic variable derived from this point cloud, essentially has/have the same distribution for the synthetic LIDAR signals as for the measured LIDAR signals.

Abstract: A computer-implemented method and system are described for training a class-conditional generative adversarial network (GAN). The discriminator is trained using a classification loss function while omitting using an adversarial loss function. Instead, if the training data has C classes, the classification loss function is formulated as a 2C-class classification problem, by which the discriminator is trained to distinguish 2 times C classes. Such trained discriminator provides an informative training signal for the generator to learn the class-conditional data synthesis by the generator. A data synthesis system and computer-implemented method are also described for synthesizing data using the generative part of the trained generative adversarial network.

Abstract: A method for operating a neural network is described comprising determining, for neural network input sensor data, neural network output data using the neural network, selecting a portion of output data points to form a region of interest and determining, for each of at least some output data points outside the region of interest, a contribution value representing a contribution of one or more input data points associated with the output data point for the neural network determining the output data point values assigned to output data points in the region of interest.

Abstract: A training system for training a generator neural network arranged to transform measured sensor data into generated sensor data. The generator network is arranged to receive as input sensor data and a transformation goal selected from a plurality of transformation goals and is arranged to transform the sensor data according to the transformation goal.