Abstract: Apparatuses, systems, and techniques to construct a neural network architecture. In at least one embodiment, candidate neural components may be selected for the neural network by jointly updating performance metric masks attached to these candidate neural components and a union neural network comprising all candidate neural components.

Abstract: Apparatuses, systems, and techniques to generate labels for images. In at least one embodiment, one or more labels of one or more images are generated based, at least in part, on an amount by which the one or more images were modified.

Abstract: In various examples, a technique for processing sensor data includes generating, using a machine learning model and based on a first sensor data instance, a first set of confidences for a set of output types and a first adversarial confidence that represents a likelihood that the first sensor data instance is adversarial. The technique also includes determining that the first sensor data instance is adversarial based on the first adversarial confidence. The technique further includes transmitting a first indication that the first sensor data instance is adversarial to one or more downstream components such that the one or more downstream components perform one or more operations based at least on the indication.

Abstract: Apparatuses, systems, and methods to enable matrix multiplication acceleration by modifying an input to apply sparsity through sparse activation filtering. In at least one embodiment, a neural network modifies pixels within an image through sparse activation filtering to enable use of one or more matrix multiplication acceleration units to perform a sparse patch embedding operation.

Abstract: A wavelength conversion unit arrangement includes a carrier and a wavelength conversion unit. The wavelength conversion unit includes a wavelength conversion layer and a filter layer, and the filter layer attaches the wavelength conversion unit to the carrier. The filter layer has a first surface facing the carrier and a second surface opposite the first surface, and the first surface and the second surface have different textures.

Abstract: Exemplary systems and methods are disclosed for allocating resources in a breeding pipeline to multiple origins. One exemplary method includes accessing a data structure including data representative of multiple origins, in which the data includes, for each of the multiple origins, a trait performance expression or genotypic component information. The exemplary method further includes determining a resource allocation, which allocates n resources among the multiple origins based on a probability associated with the trait performance expressions and/or the genotypic components for the origins, and then allocating the n resources in the breeding pipeline for the multiple origins, based on the determined resource allocation.

Abstract: A display device includes a first light-emitting module and a second light-emitting module. Each light-emitting module has a substrate, a plurality of LED dies arranged on the substrate, a reflective layer on the substrate, and a light-transmissive layer. The light-transmissive layer covers the substrate, the plurality of LED dies, and the reflective layer. Both the light-transmissive layer of the first module and the light-transmissive layer of the second module have rough uppermost surfaces. The first light-emitting module has a first reflectivity, the second light-emitting module has a second reflectivity, and a standard deviation between the first reflectivity and the second reflectivity is not greater than 0.5.

Abstract: Computing devices, computer-readable storage media, and computer-implemented methods are disclosed for prediction of capacity. In a central tier, central-tier benchmark values are generated from benchmark testing performed on different test configurations in a reference execution environment. In a deployment tier, deployment-tier benchmark values are generated from benchmark testing performed on a baseline deployed configuration in many execution environments. A sizing model is learned from the central-tier benchmark values to predict execution platform requirements given a set of workload input parameters. A performance model is learned from the deployment-tier and the central-tier benchmark values to predict a performance delta value reflecting relative performance between a particular execution environment and the reference execution environment. The performance delta value is used to adjust predicted execution platform requirements to tailor the prediction to a particular execution environment.

Abstract: Apparatuses, systems, and techniques to compress neural networks. In at least one embodiment, one or more first neural networks are used to cause one or more compressed neural networks to be selected based, at least in part, on accuracy and performance of the one or more compressed neural networks.

Abstract: Apparatuses, systems, and techniques to process neural networks. In at least one embodiment, neural network graph data is organized for processing. In at least one embodiment, for example, neural network graph data is organized based, at least in part, on one or more sparsity constraints.

Abstract: Apparatuses, systems, and techniques are presented to generate digital reconstructions of physical objects. In at least one embodiment, one or more first neural networks are used to generate a three-dimensional (3D) model having a first level of detail, and one or more second neural networks are used to modify the 3D model to have a second level of detail.

Abstract: Apparatuses, systems, and techniques to evaluate neural networks. In at least one embodiment, neural networks are evaluated using one or more other neural networks. In at least one embodiment, two or more neural networks are caused to generate consistent results from first input information and caused to generate inconsistent results from second input information.

Abstract: Exemplary systems and methods are disclosed for allocating resources in a breeding pipeline to multiple origins. One exemplary method includes accessing a data structure including data representative of multiple origins, in which the data includes, for each of the multiple origins, a trait performance expression or genotypic component information. The exemplary method further includes determining a resource allocation, which allocates n resources among the multiple origins based on a probability associated with the trait performance expressions and/or the genotypic components for the origins, and then allocating the n resources in the breeding pipeline for the multiple origins, based on the determined resource allocation.

Abstract: Apparatuses, systems, and techniques to generate an image. In at least one embodiment, one or more neural networks are to generate a second image based, at least in part, on a first image and information indicating zero or more differences between the first and second image.

Abstract: Apparatuses, systems, and techniques to modify tensors based on processor requirements. In at least one embodiment, input tensors and weight tensors are modified to meet processing resource requirements.

Abstract: A light-emitting device includes a circuit carrier board having a short side and a long side, a plurality of light-emitting units on the circuit carrier board for emitting three or more color lights, and a light-transmitting glue layer on the circuit carrier board and covering the plurality of light-emitting units. The short side is shorter than the long side. The plurality of light-emitting units include a first light-emitting unit. The first light-emitting unit has a light exit surface, a first sidewall, and a second sidewall. The first sidewall faces the short side and has a first included angle with the light exit surface, and the second sidewall faces the long side and has a second included angle with the light exit surface. The first included angle is between 85 to 95 degrees, and the second included angle is less than 85 degrees or greater than 105 degrees.

Abstract: Exemplary systems and methods are disclosed for allocating resources in a breeding pipeline to multiple origins. One exemplary method includes accessing a data structure including data representative of multiple origins, in which the data includes, for each of the multiple origins, a trait performance expression or genotypic component information. The exemplary method further includes determining a resource allocation, which allocates n resources among the multiple origins based on a probability associated with the trait performance expressions and/or the genotypic components for the origins, and then allocating the n resources in the breeding pipeline for the multiple origins, based on the determined resource allocation.

Abstract: A display device includes a first light-emitting module and a second light-emitting module. Each light-emitting module has a substrate, a plurality of LED dies arranged on the substrate, a reflective layer on the substrate, and a light-transmissive layer. The light-transmissive layer covers the substrate, the plurality of LED dies, and the reflective layer. Both the light-transmissive layer of the first module and the light-transmissive layer of the second module have rough uppermost surfaces. The first light-emitting module has a first reflectivity, the second light-emitting module has a second reflectivity, and a standard deviation between the first reflectivity and the second reflectivity is not greater than 0.5.

Abstract: Exemplary systems and methods are disclosed for allocating resources in a breeding pipeline to multiple origins. One exemplary method includes accessing a data structure including data representative of multiple origins, in which the data includes, for each of the multiple origins, a trait performance expression or genotypic component information. The exemplary method further includes determining a resource allocation, which allocates n resources among the multiple origins based on a probability associated with the trait performance expressions and/or the genotypic components for the origins, and then allocating the n resources in the breeding pipeline for the multiple origins, based on the determined resource allocation.

Abstract: Methods, apparatus, systems and articles of manufacture for generating a three-dimensional (3D) model for 3D scene reconstruction are disclosed. An example apparatus includes a 3D scene generator to generate a 3D model for digital image scene reconstruction based on a trained generative model and a digital image captured in a real environment. An image simulator is to generate a simulated image based on the 3D model, the simulated image corresponding to the captured image. A discriminator is to apply a discriminative model to the simulated image to determine whether the simulated image is simulated.