Abstract: Computing devices, such as mobile computing devices, have access to one or more image sensors that can capture images with multiple subjects. Some of these subjects may be known to the user capturing an image with the image sensor. The user may prefer to have the captured image data be optimized around the known subjects. Low-power, fast-response machine learning logic can be configured to allow for the generation of a plurality of inference data. This inference data can be utilized along with other sensor data, such as a motion sensor, for the generation of one or more image sensor configuration changes that may be implemented to optimize the subsequent capture of image data. This cycle of image data analysis, image sensor optimization, and subsequent capture can continue multiple times until a threshold of optimization or time is met. The captured image data optimized around the known subjects is then stored.

Abstract: Profile based services and content can assist with the backward and forward compatibility of conventional service layers or the like. Using a profile to manage content and access to services offered may reduce the overhead or processing associated with providing those value added services by service layers.

Abstract: The present disclosure generally relate to spin-orbit torque (SOT) devices comprising a topological insulator (TI) modulation layer. The TI modulation layer comprises a plurality of bismuth or bismuth-rich composition modulation layers, a plurality of TI lamellae layers comprising BiSb having a (012) crystal orientation, and a plurality of texturing layers. The TI lamellae layers comprise dopants or clusters of atoms, the clusters of atoms comprising a carbide, a nitride, an oxide, or a composite ceramic material. The clusters of atoms are configured to have a grain boundary glass forming temperature of less than about 400° C. Doping the TI lamellae layers comprising BiSb having a (012) crystal orientation with clusters of atoms comprising a carbide, a nitride, an oxide, or a composite ceramic material enable the SOT MTJ device to operate at higher temperatures while inhibiting migration of Sb from the BiSb of the TI lamellae layers.

Abstract: Techniques are provided for customizing or fine-tuning a pre-trained version of a machine-learning model that includes multiple layers and is configured to process audio or textual language input. Each of the multiple layers is configured with a plurality of layer-specific pre-trained parameter values corresponding to a plurality of parameters, and each of the multiple layers is configured to implement multi-head attention. An incomplete subset of the multiple layers is identified for which corresponding layer-specific pre-trained parameter values are to be fine-tuned using a client data set. The machine-learning model is fine-tuned using the client data set to generate an updated version of the machine-learning model, where the layer-specific pre-trained parameter values configured for each layer of one of more of the multiple layers not included in the incomplete subset are frozen during the fine-tuning. Use of the updated version of the machine-learning model is facilitated.

Abstract: A method of predicting an output of oil and gas production in a hydrocarbon reservoir of a current and future producing well using a neural network model, includes receiving a data set comprising a plurality of parameters of the hydrocarbon reservoir at a wellsite. The method also includes using the data set to generate a plurality of simulation curves of the hydrocarbon reservoir, each parameter of the plurality of parameters has a range, and the range is adjustable, and the wellsite includes a wellbore penetrating a subterranean formation to extract reserves from the hydrocarbon reservoir. The method also includes performing a simulation, based on the range of each said parameter of the plurality of parameters, of the hydrocarbon reservoir. The method also includes downloading the plurality of simulation curves into a local server to prepare training data for training the neural network model.

Abstract: Containers with one or more tray retention portions and a support stand are described. For example, a container can include: a substantially hollow body including a plurality of body walls extending vertically from a bottom portion; an extension portion rotatably coupled to an outward-facing surface of a body wall among the plurality of body walls, the extension portion configured to rotate to an active position in which the extension portion extends outwardly from the body wall; and a support stand having a proximal end coupled to the outward-facing surface of the body wall and a distal end, the support stand configured to move between at least a stowed position in which the support stand is positioned against or proximate to the outward-facing surface of the body wall and a support position in which the support stand extends outwardly from the body wall.

Abstract: Systems and methods are described herein to automate managing of service layer operations comprised of multiple elementary operations and offloading the burden of performing such multi-step operations from a requesting entity to the service layer. A Request Abstraction Service (RAS) is described herein for the autonomous execution of such multi-step operations. Methods and apparatuses are also described herein for a service layer framework for integrating generic and functional user interfaces as services managed by the SL on behalf of requesting entities.

Abstract: A method of providing prioritized call route lists to network elements of communication service carriers. The method comprises building prioritized route lists by a centralized routing function (CRF) platform by processing prioritized route list transactions built by a route generation application based on key performance indicators, receiving a request for a prioritized route list by the CRF platform from a session border controller (SBC) controlled by the first communication service carrier, where the request comprises an identity of a second communication service carrier and a destination telephone number, determining a subscriber service level associated with the second communication service carrier by the CRF platform, based on the determined subscriber service level, looking up a prioritized route list by the CRF platform based on the destination telephone number, and sending the looked up prioritized route list by the CRF platform to the SBC.

Abstract: A method of triggering a security alert using a tracking device, including detecting motion using a motion sensor of the tracking device, determining that the detected motion meets a motion threshold, detecting one or more wireless access points in proximity to the tracking device, determining that the one or more wireless access points do not correspond to a recognized wireless access point, and in accordance with the determination that the one or more detected wireless access points do not correspond to a recognized wireless access point, triggering a security alert.

Abstract: A motor housing includes a bearing holder that holds a first bearing and accommodates a rotor and a stator. A second bearing is disposed inside a motor shaft at one axial end portion of the motor shaft and has lower electric resistance than the first bearing. The conductive member electrically connects the second bearing to a motor housing. The radially outer end portion of the second bearing is in contact with the inner peripheral surface of the motor shaft. The contact portion of the conductive member is in contact with the radially inner end portion of the second bearing. The fixed portion is fixed to the motor housing radially outward of the motor shaft. A bridge portion connects the contact portion and the fixed portion.

Abstract: Techniques are disclosed herein for focused training of language models and end-to-end hypertuning of the framework. In one aspect, a method is provided that includes obtaining a machine learning model pre-trained for language modeling, and post-training the machine learning model for various tasks to generate a focused machine learning model. The post-training includes: (i) training the machine learning model on an unlabeled set of training data pertaining to a task that the machine learning model was pre-trained for as part of the language modeling, and the unlabeled set of training data is obtained with respect to a target domain, a target task, or a target language, and (ii) training the machine learning model on a labeled set of training data that pertains to another task that is an auxiliary task related to a downstream task to be performed using the machine learning model or output from the machine learning model.

Abstract: A fog service layer architecture is disclosed using hierarchical fog node deployment including the co-existence and interactions of the fog node with a cloud node. The architecture also includes a list of functions, capabilities or services that are hosted in each fog node. One or more fog management procedures may be run between fog nodes (or between fogs and the cloud) and may comprise a fog capability discovery procedure, a fog connection verification procedure, and a fog capability status report procedure. In addition, fog nodes may be configured to interact with each other to get particular services using one or more fog service procedures described herein.

Abstract: The present disclosure generally relates to spin-orbital torque (SOT) differential reader designs. The SOT differential reader is a multi-terminal device that comprises a first shield, a first spin hall effect layer, a first free layer, a gap layer, a second spin hall effect layer, a second free layer, and a second shield. The gap layer is disposed between the first spin hall effect layer and the second spin hall effect layer. Electrical lead connections are located about the first spin hall effect layer, the second spin hall effect layer, the gap layer, the first shield, and/or the second shield. The electrical lead connections facilitate the flow of current and/or voltage from a negative lead to a positive lead. The positioning of the electrical lead connections and the positioning of the SOT differential layers improves reader resolution without decreasing the shield-to-shield spacing (i.e., read-gap).

Abstract: In a method of operating a computer system, an instruction loop is executed by a processor in which each iteration of the instruction loop accesses a current data vector and an associated current vector predicate. The instruction loop is repeated when the current vector predicate indicates the current data vector contains at least one valid data element and the instruction loop is exited when the current vector predicate indicates the current data vector contains no valid data elements.

Abstract: A neural processing unit of a binarized neural network (BNN) as a hardware accelerator is provided, for the purpose of reducing hardware resource demand and electricity consumption while maintaining acceptable output precision. The neural processing unit may include: a first block configured to perform convolution by using a binarized feature map with a binarized weight; and a second block configured to perform batch-normalization on an output of the first block. A register having a particular size may be disposed between the first block and the second block. Each of the first block and the second block may include one or more processing engines. The one or more processing engines may be connected in a form of pipeline.

Abstract: A digital signal processor having at least one streaming address generator, each with dedicated hardware, for generating addresses for writing multi-dimensional streaming data that comprises a plurality of elements. Each at least one streaming address generator is configured to generate a plurality of offsets to address the streaming data, and each of the plurality of offsets corresponds to a respective one of the plurality of elements. The address of each of the plurality of elements is the respective one of the plurality of offsets combined with a base address.

Abstract: A method of analyzing a welded connection during laser welding of workpieces includes acquiring a first measurement signal for a process radiation generated during laser welding, acquiring a second measurement signal for a laser radiation reflected by the workpieces, determining whether there is a gap between the workpieces based on the first measurement signal, and when it is determined that there is a gap, determining based on the second measurement signal whether there is a welded connection.

Abstract: A device for stabilizing an image-capturing device against trembling and shaking includes a chip, an image sensor, a first connector, and a second connector. The image sensor collects light and converts same into an electrical signal. The first connector is used for receiving the electrical signal. The second connector is used for electrically coupling the first connector and the chip, wherein the second connector transmits the electrical signal to the chip. The stabilizing device is elastically suspended in place and tremors and shakes on the part of the stabilizing device are cancelled as they occur by reverse displacement based on magnetism and as controlled by Hall effect sensors. The chip can output control signal to the second connector, the second connector transmits the control signal to the first connector, the first connector transmits the control signal to the image sensor to control state of the image sensor.

Abstract: A semiconductor package assembly includes a first mounting surface of a package substrate that faces a second mounting surface of a printed circuit board. A first structural element bond pad is mounted to the first mounting surface. A second structural element bond pad is mounted to the second mounting surface, and the first and second structural element bond pads are aligned with each other. A structural element is interconnected with a first solder joint to the first structural element bond pad and interconnected with a second solder joint to the second structural element bond pad. The structural element extends between the first and second structural element bond pads to absorb mechanical shock when a compressive force pushes one of the first and second mounting surfaces toward the other.

Abstract: In one example in accordance with the present disclosure, a system is described. The system includes a reader to extract cleaning instructions associated with a three-dimensional (3D) printed object. The cleaning instructions include a termination condition to indicate when object cleaning is complete. The system also includes a controller to instruct at least one cleaning device to clean the 3D printed object based on the cleaning instructions. A measurement system of the system determines when the termination condition is met.