Abstract: A computer-implemented method for training a normalizing flow. The normalizing flow predicts a first density value based on a first input image. The first density value characterizes a likelihood of the first input image to occur. The first density value is predicted based on an intermediate output of a first convolutional layer of the normalizing flow. The intermediate output is determined based on a plurality of weights of the first convolutional layer. The method for training includes: determining a second input image; determining an output, wherein the output is determined by providing the second input image to the normalizing flow and providing an output of the normalizing flow as output; determining a second density value based on the output tensor and on the plurality of weights; determining a natural gradient of the plurality of weights with respect to the second density value; adapting the weights according to the natural gradient.

Abstract: A helmet has an outer shell, an impact absorbing liner inside the outer shell to receive loads from the outer shell and an inner liner disposed inside the impact absorbing liner, with the inner liner configured to slide relative to the impact absorbing liner. The inner liner defines a number of apertures and a deflector inside each aperture. Each deflector has a body that is connectable to the impact absorbing liner; a peripheral border that is connectable to the aperture in which it is received; and a number of deformable spokes extending between the body and the border.

Abstract: A helmet has an outer shell, an impact absorbing liner inside the outer shell to receive loads from the outer shell and an inner liner disposed inside the impact absorbing liner, with the inner liner configured to slide relative to the impact absorbing liner. The inner liner defines a number of apertures and a deflector inside each aperture. Each deflector has a body that is connectable to the impact absorbing liner; a peripheral border that is connectable to the aperture in which it is received; and a number of deformable spokes extending between the body and the border.

Abstract: A computer-implemented method for training a normalizing flow. The normalizing flow is configured to determine a first output signal characterizing a likelihood or a log-likelihood of an input signal. The normalizing flow includes at least one first layer which includes trainable parameters. A layer input to the first layer is based on the input signal and the first output signal is based on a layer output of the first layer. The training includes: determining at least one training input signal; determining a training output signal for each training input signal using the normalizing flow; determining a first loss value which is based on a likelihood or a log-likelihood of the at least one determined training output signal with respect to a predefined probability distribution; determining an approximation of a gradient of the trainable parameters; updating the trainable parameters of the first layer based on the approximation of the gradient.

Abstract: A computer-implemented method for training a normalizing flow. The normalizing flow predicts a first density value based on a first input image. The first density value characterizes a likelihood of the first input image to occur. The first density value is predicted based on an intermediate output of a first convolutional layer of the normalizing flow. The intermediate output is determined based on a plurality of weights of the first convolutional layer. The method for training includes: determining a second input image; determining an output, wherein the output is determined by providing the second input image to the normalizing flow and providing an output of the normalizing flow as output; determining a second density value based on the output tensor and on the plurality of weights; determining a natural gradient of the plurality of weights with respect to the second density value; adapting the weights according to the natural gradient.

Abstract: An item of footwear such as a boot (10) includes a stiff lower part (12) that extends around part of a wearer's foot, a stiff upper part (16) that extends around the wearer's lower leg, and one or more limit elements (66,72,80,88) that extend between the lower part (12) and the upper part (16). The lower part (12) and upper part (16) are movable relative to each other when the ankle moves and the relative movement is inhibited selectively by the limit elements (66,72,80,88). The limit elements (66,72,80,88) are displaceable between a free condition in which the lower part (12) and the upper part (16) are movable relative to each other, and lock condition in which the relative movement between the lower part (12) and the upper part (16) is inhibited.

Abstract: A helmet has an outer shell, an impact absorbing liner inside the outer shell to receive loads from the outer shell and an inner liner disposed inside the impact absorbing liner, with the inner liner configured to slide relative to the impact absorbing liner. The inner liner defines a number of apertures and a deflector inside each aperture. Each deflector has a body that is connectable to the impact absorbing liner; a peripheral border that is connectable to the aperture in which it is received; and a number of deformable spokes extending between the body and the border.

Abstract: A helmet comprises an outer shell, an impact absorbing liner inside the outer shell to receive loads from the outer shell and an inner liner disposed inside the impact absorbing liner, with the inner liner configured to slide relative to the impact absorbing liner. The inner liner defines a number of apertures and a deflector inside each aperture. Each deflector comprises: a body that is connectable to the impact absorbing liner; a peripheral border that is connectable to the aperture in which it is received; and a number of deformable spokes extending between the body and the border.

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 device and a method for generating a compressed network from a trained neural network are provided. The method includes: a model generating a compressing map from first training data, the compressing map representing the impact of model components of the model to first output data in response to the first training data; generating a compressed network by compressing the trained neural network in accordance with the compressing map; the trained neural network generating trained network output data in response to second training data; the compressed network generating compressed network output data in response to the second training data; training the model by comparing the trained network output data with the compressed network output data.

Abstract: A helmet comprises an outer shell, an impact absorbing liner inside the outer shell to receive loads from the outer shell and an inner liner disposed inside the impact absorbing liner, with the inner liner configured to slide relative to the impact absorbing liner. The inner liner defines a number of apertures and a deflector inside each aperture. Each deflector comprises: a body that is connectable to the impact absorbing liner; a peripheral border that is connectable to the aperture in which it is received; and a number of deformable spokes extending between the body and the border.

Abstract: An item of footwear such as a boot (10) includes a stiff lower part (12) that extends around part of a wearer's foot, a stiff upper part (16) that extends around the wearer's lower leg, and one or more limit elements (66,72,80,88) that extend between the lower part (12) and the upper part (16). The lower part (12) and upper part (16) are movable relative to each other when the ankle moves and the relative movement is inhibited selectively by the limit elements (66,72,80,88). The limit elements (66,72,80,88) are displaceable between a free condition in which the lower part (12) and the upper part (16) are movable relative to each other, and lock condition in which the relative movement between the lower part (12) and the upper part (16) is inhibited.

Abstract: A helmet comprises an outer shell, an impact absorbing liner (10) inside the outer shell to receive loads from the outer shell and an inner liner (12) disposed inside the impact absorbing liner (10), with the inner liner (12) configured to slide relative to the impact absorbing liner (10). The inner liner (12) defines a number of apertures (16) and a deflector (14) inside each aperture. Each deflector (14) comprises: a body (20) that is connectable to the impact absorbing liner (12); a peripheral border (18) that is connectable to the aperture (16) in which it is received; and a number of deformable spokes (22) extending between the body (20) and the border (18).

Abstract: Protective wear such as a glove comprises a textile fabric substrate with elements of a synthetic plastics material that hardens on being subjected to impact, injection molded onto the fabric. The glove has a row of elements along each digit and some of the elements have recesses in them to provide space for the knuckles and finger joints of the hand.

Abstract: A knee brace (10) according to the invention includes an upper leg formation (12), a lower leg formation (14) and a hinge mechanism (16) that is disposed entirely on the lateral side of the knee. The knee brace (10) has pressure formations (20, 24,62, 28) pressing on three pressure points: on the lateral aspect of the wearer's upper leg, on the lateral aspect of the wearer's lower leg, and against the knee complex on the medial aspect of the distal extremity of the femur. The three pressure formations are connected to one another by a stiff structure (18, 16) which includes the hinge mechanism (16).

Abstract: The invention relates to the use of alpha 1-antichymotrypsin (ACT) polypeptides according to SEQ ID No. 1 to SEQ ID No. 4 and/or nucleic acids encoding them, or an antibody or a fragment thereof directed against the polypeptide, or of a cell which is expressing the polypeptide or a nucleic acid encoding it, for diagnosis, treatment and/or prevention of diabetes-associated and/or arterial wounds which heal poorly and for identifying pharmacologically active substances which exert an influence on the expression or function, particularly the activity of ACT.

Abstract: The invention relates to a use of alpha 1-antichymotrypsin (ACT) polypeptides, functional variants thereof and/or nucleic acids encoding them, or of a cell which is expressing an ACT polypeptide or a nucleic acid encoding it in combination with alpha-1-antitrypsin (AAT) polypeptides, functional variants thereof or nucleic acids encoding them, or of a cell which is expressing an AAT polypeptide, or a nucleic acid encoding it.

Abstract: Methods of promoting wound healing or treating a wound healing disorder in a patient and methods of preventing or treating a skin disorder in a patient are disclosed. As an active ingredient Eps 8 is disclosed.