Abstract: A graph neural network (GNN) model is used in a scheduling process for compiling a deep neural network (DNN). The DNN, and parameter options for scheduling the DNN, are represented as a graph, and the GNN predicts a set of parameters that is expected to have a low cost. Using the GNN-based model, a compiler can produce a schedule for compiling the DNN in a relatively short and predictable amount of time, even for DNNs with many layers and/or many parameter options. For example, the GNN-based model reduces the overhead of exploring every parameter combination and does not exclude combinations from consideration like prior heuristic-based approaches.

Abstract: A machine learning system is provided to enhance various aspects of machine learning models. In some aspects. a substantially photorealistic three-dimensional (3D) graphical model of an object is accessed and a set of training images of the 3D graphical mode are generated, the set of training images generated to add imperfections and degrade photorealistic quality of the training images. The set of training images are provided as training data to train an artificial neural network.

Abstract: A machine learning system is provided to enhance various aspects of machine learning models. In some aspects, a substantially photorealistic three-dimensional (3D) graphical model of an object is accessed and a set of training images of the 3D graphical mode are generated, the set of training images generated to add imperfections and degrade photorealistic quality of the training images. The set of training images are provided as training data to train an artificial neural network.

Abstract: Deep learning operations (e.g., transposed convolution, resized convolution, dilated convolution, etc.) may be performed with sparsity maps and storage pointers. A deep learning operation has a tensor, which can be used to generate an upsampled tensor by adding new data elements (e.g., zeros) into the tensor. One or more sparsity maps may be generated based on one or more parameters of the first deep learning operation. The sparsity map may include elements indicating whether a data element in the upsampled tensor is a data element in the tensor or is a new data element. One or more storage pointers may be generated. A storage pointer may indicate a location (e.g., a memory address) where one or more data elements of the tensor are stored in a memory. An output of the deep learning operation may be performed using data elements in the tensor, the sparsity maps, and the storage pointers.

Abstract: A deconvolution can be decomposed into multiple convolutions. Results of the convolutions constitute an output of the deconvolution. Zeros may be added to an input tensor of the deconvolution to generate an upsampled input tensor. Subtensors having the same size as the kernel of the deconvolution may be identified from the upsampled input tensor. A subtensor may include one or more input activations and one or more zeros. Subtensors having same distribution patterns of input activations may be used to generate a reduced kernel. The reduced kernel includes a subset of the kernel. The position of a weight in the reduced kernel may be the same as the positions of an input activation in the subtensor. Multiple reduced kernels may be generated based on multiple subtensors having different distribution patterns of activations. Each of the convolutions may use the input tensor and a different one of the reduced kernels.

Abstract: Technology for estimating neural network (NN) power profiles includes obtaining a plurality of workloads for a compiled NN model, the plurality of workloads determined for a hardware execution device, determining a hardware efficiency factor for the compiled NN model, and generating, based on the hardware efficiency factor, a power profile for the compiled NN model on one or more of a per-layer basis or a per-workload basis. The hardware efficiency factor can be determined on based on a hardware efficiency measurement and a hardware utilization measurement, and can be determined on a per-workload basis. A configuration file can be provided for generating the power profile, and an output visualization of the power profile can be generated. Further, feedback information can be generated to perform one or more of selecting a hardware device, optimizing a breakdown of workloads, optimizing a scheduling of tasks, or confirming a hardware device design.

Abstract: A particular voxel is identified within a volume and a hash table is used to obtain volumetric data describing the particular voxel within the volume. Values of x-, y- and z-coordinates in the volume associated with the particular voxel are determined an index value associated with the particular voxel is determined according to a hashing algorithm, where the index value is determined from summing weighted values of the x-, y- and z-coordinates, and the weighted values are based on a variable value corresponding to a dimension of the volume. A particular entry is identified in the hash table based on the index value, where the particular entry includes volumetric data, and the volumetric data identifies, for the particular voxel, whether the particular voxel is occupied.

Abstract: Systems, apparatuses and methods may provide for technology that determines a complexity of a task associated with a neural network workload and generates a hardware efficiency estimate for the task, wherein the hardware efficiency estimate is generated via a neural network based cost model if the complexity exceeds a threshold, and wherein the hardware efficiency estimate is generated via a cost function if the complexity does not exceed the threshold. In one example, the technology trains the neural network based cost model based on one or more of hardware profile data or register-transfer level (RTL) data.

Abstract: Methods, apparatus, systems, and articles of manufacture are disclosed for performing a machine learning operation using storage element pointers. An example computer readable medium comprises instructions that when executed, cause at least one processor to select, in response to a determination that a machine learning operation is to be performed, create first and second storage element pointers based on a type of machine learning operation to be performed, remap input tensor data of the input tensor based on the first storage element pointer without movement of the input tensor data in memory, cause execution of the machine learning operation with the remapped input tensor data to create intermediate tensor data, remap the intermediate tensor data based on the second storage element pointer without movement of the intermediate tensor data in memory, and provide the remapped intermediate tensor data as an output tensor.

Abstract: Methods, apparatus, systems, and articles of manufacture are disclosed that increase utilization of neural network (NN) accelerator circuitry for shallow layers of an NN by reformatting one or more tensors. An example apparatus includes parameter determining circuitry to determine a width of a weight kernel and to determine a depth of a first tensor. The example apparatus also includes storage control circuitry to, starting at a first XY location of the first tensor, copy one or more Z values, up to the depth of the first tensor, of consecutive XY locations that overlap the width of the weight kernel and to load the one or more Z values consecutively in a first XY location of a second tensor.

Abstract: A machine learning system is provided to enhance various aspects of machine learning models. In some aspects, a substantially photorealistic three-dimensional (3D) graphical model of an object is accessed and a set of training images of the 3D graphical mode are generated, the set of training images generated to add imperfections and degrade photorealistic quality of the training images. The set of training images are provided as training data to train an artificial neural network.

Abstract: A particular voxel is identified within a volume and a hash table is used to obtain volumetric data describing the particular voxel within the volume. Values of x-, y- and z-coordinates in the volume associated with the particular voxel are determined an index value associated with the particular voxel is determined according to a hashing algorithm, where the index value is determined from summing weighted values of the x-, y- and z-coordinates, and the weighted values are based on a variable value corresponding to a dimension of the volume. A particular entry is identified in the hash table based on the index value, where the particular entry includes volumetric data, and the volumetric data identifies, for the particular voxel, whether the particular voxel is occupied.

Abstract: Methods, apparatus, systems and articles of manufacture to reconstruct scenes using convolutional neural networks are disclosed. An example apparatus includes a sensor data acquirer to acquire ground truth data representing an environment, an environment detector to identify an environmental characteristic of the environment, a synthetic database builder to apply noise to the ground truth data to form a training set, a model builder to train a machine learning model using the training set and the ground truth data, and a model adjustor to modify the machine learning model to include residual OR-gate connections intermediate respective layers of the machine learning model. The synthetic database builder is further to store the machine learning model in association with the environmental characteristic of the environment.