Micron Technology Incorporated

Colorado State University Research Foundation

Saideep Tiku

Poorna Kale

Sudeep Pasricha

Robert Noel Bielby

Methods, systems, and devices for video stream augmentation using a deep learning device are described. A machine learning device of a vehicle may augment a video stream received from cameras of the vehicle and may output the augmented video stream to a display component of the vehicle. For example, a camera of the vehicle may record a video stream of and a sensor of the vehicle may detect information about an environment associated with the vehicle. The camera and sensor may transmit the video stream and information, respectively, to the machine learning device, which may process and modify the video stream based on parameters of the video stream and/or the information. The machine learning device may transmit the modified video streams to the display component, and the display component may display aspects of the modified video stream on a display of the vehicle, such as a rearview mirror.

Video stream augmentation using a deep learning device

In some implementations, an extended reality (XR) device may detect, using a camera of the XR device, a clothing item, wherein the clothing item is associated with an identifier. The XR device may transmit, to a server, a request that indicates the identifier. The XR device may receive, from the server, metadata associated with the clothing item, wherein the metadata is associated with the identifier. The XR device may retrieve, from the server, a three-dimensional model of a user associated with the XR device. The XR device may generate a three-dimensional model of the user wearing the clothing item using the three-dimensional model of the user and the metadata. The XR device may provide, via an interface of the XR device, the three-dimensional model of the user wearing the clothing item.

Three-dimensional models of users wearing clothing items

An artificial neural network can be allocated to memory and operated. An error can occur in the memory and/or be detected in the memory. Layers of the artificial neural network can be remapped in the memory at least partially in response to the error. Performance of the artificial neural network can be evaluated before and/or after the remapping.

Artificial neural network remapping in memory

Apparatuses and methods can be related to encoding traffic between a host and a deep learning accelerator (DLA). Traffic between a host can be encoded utilizing an autoencoder. Encoding traffic between a host and a DLA changes the bandwidth of the traffic. Changing the bandwidth of the traffic prevents the correlation between the bandwidth and the input from which the traffic is generated.

Encoded host to DLA traffic

Apparatuses and methods can be related to implementing bypass paths in an ANN. The bypass path can be used to bypass a portion of the ANN such that the ANN generates an output with a particular level of confidence while utilizing less resources than if the portion of the ANN had not been bypassed.

Artificial neural network bypass

Multiple artificial neural networks can be compiled as a single workload. A respective throughput for each of the artificial neural networks can be changed at runtime. The multiple artificial neural networks can be partially compiled individually and then later compiled just-in-time according to changing throughput demands for the artificial neural networks. The multiple artificial neural networks can be deployed on a deep learning accelerator hardware device.

Artificial neural networks on a deep learning accelerator

In some implementations, an extended reality (XR) device may receive, via an interface of the XR device, an input associated with meals of a user associated with the XR device. The XR device may determine, based on the input, a meal plan for the user of the XR device, wherein the meal plan is associated with target meals. The XR device may determine, based on recipes for the target meals, a list of food items for preparing the target meals associated with the meal plan. The XR device may provide, via the interface, the list of food items and the target meals. The XR device may provide, via the interface, an in-store navigation path to direct the user of the XR device via overlayed audio-visual cues to locations within a physical retail store to pick up the food items.

Navigation paths for directing users to food items based on meal plans

Apparatuses and methods can be related to implementing age-based network training. An artificial neural network (ANN) can be trained by introducing errors into the ANN. The errors and the quantity of errors introduced into the ANN can be based on age-based characteristics of the memory device.

Artificial neural network training in memory

In some implementations, an extended reality (XR) device may generate a first event in a virtual environment, wherein the first event is associated with a first difficulty level. The XR device may detect vital information of a user associated with the XR device, wherein the vital information is associated with the first event. The XR device may determine whether the vital information satisfies a threshold. The XR device may transmit, to a server, an indication that indicates whether the vital information satisfies the threshold. The XR device may receive, from the server, virtual environment information, wherein the virtual environment information is based on whether the vital information satisfies the threshold. The XR device may generate, based on the virtual environment information, a second event in the virtual environment, wherein the second event is associated with a second difficulty level.

Providing scenes in a virtual environment based on health data using extended reality

An exemplary radio fingerprint-based indoor localization method and system is disclosed that is resistant to spoofing or jamming attacks (e.g., at nearby radios, e.g., access points), among other types of interference. The exemplary method and system may be applied in the configuring of a secured convolutional neural network (S-CNNLOC) or secured deep neural network configured for attack-resistant fingerprint-based indoor localization.

Growing community of inventors

Fort Collins, CO, United States of America

Saideep Tiku