Abstract: A resource set which includes multiple servers with a respective plurality of training computing devices is identified for training a machine learning model. The resource set is subdivided into partition groups, such that each partition group can store a respective replica of state information of the model. The model is trained using the partition groups. The training comprises a multi-stage gathering of a portion of the state information at training computing devices of a particular partition group. Different types of communication channels between training computing devices are used in respective stages of the gathering, including inter-server communication channels in one stage and an intra-server communication channel during another stage. A trained version of the model is stored.

Abstract: An automatic loading system for screening wood shavings, wherein discharging side of a material receiving and conveying mechanism is connected to feeding side of shaving arranging and conveying mechanisms, each shaving arranging and conveying mechanism includes first slide-down board, first conveying mechanism, second slide-down board, second conveying mechanism, third slide-down board, and third conveying mechanism sequentially disposed along wood shavings conveying direction, width of the first slide-down board is gradually reduced along wood shavings conveying direction to cause its discharging side's width corresponds to wood shavings'width, a baffle board is disposed above a feeding end of the second conveying mechanism with a set gap, and a feeding end of the third slide-down board locates below a discharging end of the second conveying mechanism at a first set distance, and a discharging end located above a feeding end of the third conveying mechanism at a second set distance.

Abstract: A system for calibrating the multi-element antenna can include a computer including a processor and memory, wherein the memory stores instructions executable by the processor to detect a location of a laser signal incident on a receiving surface, the laser signal having been reflected from a reflector positioned over an area of a multi-element antenna. The instructions may additionally include instructions to compute a differential phase of an element of the multi-element antenna based on the detected location of the incident laser signal.

Abstract: Novel processes are provided for the cooling of light hydrocarbon cracker effluent streams using Liquefied Natural Gas (LNG). The effluent streams are passed through LNG cold boxes where they are cooled by LNG. The cooling takes place as a result of direct or indirect heat transfer. As the LNG heats up from this process it can be fed into a light hydrocarbon cracker. The use of the present processes offers significant economic and environmental benefits over the typical industry practice of compression and cooling using C2/C3 refrigerants.

Abstract: An information transfer method, apparatus, and system are disclosed. The method performed by an IMS application server includes: receiving an SIP message from an IMS communication control network element, where the SIP message is used to manage a communication session between a first communication device and a second communication device, and the SIP message includes a charging identifier of the communication session; and sending the charging identifier to a data channel control network element, where the charging identifier is used to associate charging information of a data channel application with charging information of the communication session. In this way, the data channel control network element subsequently sends the charging identifier to a charging processing device. This helps the charging processing device perform associated charging on the charging information of the communication session and the charging information of the data channel application.

Abstract: Tuning result records indicating tuned schedules for machine learning tuning tasks are stored at a data store accessed from a tuning service. A given schedule indicates at least an order in which sub-operations of a tuning task are to be executed. In response to determining that the data store does not include a result record whose tuning task meets a similarity criterion to a tuning task determined from a tuning request, performance predictions for a set of candidate schedules are obtained using a performance prediction model, without running code which executes the candidate schedules. In a response to the tuning request, executable code corresponding to a preferred schedule identified from the candidate schedules using the performance predictions is included.

Abstract: A resource set which includes multiple servers with a respective plurality of training computing devices is identified for training a machine learning model. The resource set is subdivided into partition groups, such that each partition group can store a respective replica of state information of the model. The model is trained using the partition groups. The training comprises a multi-stage gathering of a portion of the state information at training computing devices of a particular partition group. Different types of communication channels between training computing devices are used in respective stages of the gathering, including inter-server communication channels in one stage and an intra-server communication channel during another stage. A trained version of the model is stored.

Abstract: An alloy may include iron (Fe), manganese (Mn) and aluminum (Al). The Fe may include the range 1% to 5.8% by weight of the alloy, the Mn may include the range 1.2% to 9.1% by weight of the alloy. The alloy may further include one or more of silicon (Si), nickel (Ni) and zirconium (Zr). Also, an alloy may include copper (Cu), magnesium (Mg), Zr and Al, wherein the Cu may include the range 0.8% to 5.1% by weight of the alloy, the Mg may include the range 0.5% to 3.9% by weight of the alloy and the Zr may include the range 0.3% to 10% by weight of the alloy, and the alloy may further include one or more of manganese, lithium, titanium, silicon, iron and nickel.

Abstract: This application provides an XR object rendering method, related apparatus and system. The method includes: determining or accepting, by a terminal device, a rendering division which means that a portion of to-be-rendered XR objects in an XR call of the terminal device are to be rendered by a network-side rendering device, and the other portion of to-be-rendered XR objects in the XR call of the terminal device are to be rendered by the terminal device; and rendering, by the terminal device, the to-be-rendered XR object that is to be rendered by the terminal device in the XR call, and sending, to the network-side rendering device, information about the to-be-rendered XR object that is to be rendered by the network-side rendering device in the XR call.

Abstract: Techniques for code-free automated machine learning (ML) are described. Users can train high-quality ML models and pipelines without necessarily needing to write code by providing a training dataset to a code-free machine learning service. The service may deploy an ML orchestration function and a storage location on behalf of a user. When a modification is made to the storage bucket, such as by the user providing a training dataset, the orchestration function is invoked and can automatically initiate an AutoML process using at least the training data to train multiple ML model variants. The resultant ML model(s) and associated metrics can be provided to the user, deployed behind an endpoint, and/or used to generate inferences.

Abstract: A system for calibrating the multi-element antenna can include a computer including a processor and memory, wherein the memory stores instructions executable by the processor to detect a location of a laser signal incident on a receiving surface, the laser signal having been reflected from a reflector positioned over an area of a multi-element antenna. The instructions may additionally include instructions to compute a differential phase of an element of the multi-element antenna based on the detected location of the incident laser signal.

Abstract: A V-shaped wheel spoke made of carbon fiber material is provided. The bottom part of the spoke is provided with a first through hole for placing a bolt when the spoke is connected with a wheel rim. The spoke has two legs, the top portion of each of the two legs is provided with a second through hole for placing bolt when the two legs are connected with flanges at left and right sides of a hub. With this design, the conventional connection structure consisting of spokes and nipples has been abandoned, and bolts or fastening structures are used for connection. The two legs of the spoke are respectively connected to the flanges at the left and right sides of the hub with bolts or fastening structures, such that the overall structure forms a stable triangle which greatly improves the strength and stiffness of the wheel.

Abstract: Current thermostat data including room temperature readings, thermostat control commands, and user input temperature set point changes is collected. The current thermostat data is sent from a thermostat to a cloud server that creates and continually updates a first server machine learning model trained using previous and the current thermostat data, a second thermostat machine learning model is created for sending control commands to the thermostat using the previous and the current thermostat data, and the second thermostat machine learning model is updated. The second thermostat machine learning model is updated by sending updated model parameters from the first server machine learning model to the second thermostat machine learning model. One or more new thermostat commands determined from the current thermostat data are received from the second thermostat machine learning model and the one or more new thermostat commands are executed.

Abstract: A method and an apparatus for forming powder. The formed powder may include an alloy of powder that can be used in additively manufacturing and powder metallurgy applications to create structures. The method and apparatus may deliver a source material having a first material composition, melt the source material to form a molten source material, vibrate a substate structure, the substrate structure including a substrate material having a substrate material composition, apply the molten source material to the vibrating substrate structure to obtain a powder, where a portion of the substrate material is selectively added to the molten source material such that the powder has a second material composition different than the first material composition, and control the second material composition of the powder based on the first material composition and the substrate material composition.

Abstract: Alloyed metals, and techniques for creating parts from alloyed metals, are disclosed. An apparatus in accordance with an aspect of the present disclosure comprises an alloy. An additive manufacturing alloy in accordance with the present disclosure may comprise magnesium (Mg) that is between 2.0 and 5.3% by weight, manganese (Mn) that is between 0.01 and 4.0% by weight, silicon (Si) that is between 0.1 and 1.5% by weight, zirconium (Zr) that is between 0.01 and 2.0% by weight, and aluminum (Al). In some cases, the alloy such as described in the previous sentence might not include Mg.

Abstract: A printer and methods for additive manufacturing a build piece may include a camera and an optical spectrometer obtaining spectral information and optical information from a region of melted material to determine a defect condition based on an evaluation of processed spectral or optical information. A processor or a computer may process the obtained and optical information and determine a defect condition during the additively manufacturing process. The obtained spectral and optical information may be of the region of the melted material, a melt pool and a mushy zone. The printer and method may include a controller configured to modify a process parameter to shape the weld pool to obtain a desired effective absorptivity of a portion of the weld pool, e.g., to increase the effective absorptivity relative to an absorptivity of a surface of the powder or material deposited by the depositor and to maintain an acceptable temperature of the weld pool during the additively manufacturing process.

Abstract: A two-stage machine learning model is used to for categorization of a dataset, such as transactions. A plurality of complementary base machine learning models are used to generate initial inference results and associated measures of inference confidence from the dataset, which are collected as a meta dataset. Each of the complementary models is associated with a different part of the dataset in which it has a higher accuracy in that part than the other models. The meta dataset is provided as input to a meta machine learning model, which is trained to produce a final inference result, and a confidence score model, which is trained to produce a confidence score associated with the final inference result.

Abstract: A method and an apparatus for forming powder. The formed powder may include an alloy of powder that can be used in additively manufacturing and powder metallurgy applications to create structures. The method and apparatus may deliver a source material having a first material composition, melt the source material to form a molten source material, vibrate a substate structure, the substrate structure including a substrate material having a substrate material composition, apply the molten source material to the vibrating substrate structure to obtain a powder, where a portion of the substrate material is selectively added to the molten source material such that the powder has a second material composition different than the first material composition, and control the second material composition of the powder based on the first material composition and the substrate material composition.

Abstract: Alloyed metals, and techniques for creating parts from alloyed metals, are disclosed. An apparatus in accordance with an aspect of the present disclosure comprises an alloy. An additive manufacturing alloy in accordance with the present disclosure may comprise magnesium (Mg) that is between 2.0 and 5.3% by weight, manganese (Mn) that is between 0.01 and 4.0% by weight, silicon (Si) that is between 0.1 and 1.5% by weight, zirconium (Zr) that is between 0.01 and 2.0% by weight, and aluminum (Al). In some cases, the alloy such as described in the previous sentence might not include Mg.

Abstract: Techniques for optimizing and deploying convolutional neural network (CNN) machine learning models for inference using integrated graphics processing units are described. A model compilation system optimizes CNN models using optimized vision-specific operators as well as both graph-level tuning and tensor-level tuning to explore the optimization space for achieving heightened performance. The model compilation system may also implement a heuristic-based two-stage technique for falling back certain operators of CNN models to use CPUs when needed or otherwise beneficial.