Abstract: At least one application runs on a hardware platform that includes a plurality of coprocessors, each of which has a respective internal memory space. An intermediate software layer (MVL) is transparent to the application and intercepts calls for coprocessor use. If the data corresponding to an application's call, or separate calls from different entities (including different applications) to the same coprocessor, to the API of a target coprocessor, cannot be stored within the available internal memory space of the target coprocessor, but comprises data subsets that individually can, the MVL intercepts the call response to the application/entities and indicates that the target coprocessor can handle the request. The MVL then transfers the data subsets to the target coprocessor as needed by the corresponding kernel(s) and swaps out each data subset to the internal memory of another coprocessor to make room for subsequently needed data subsets.

Abstract: An embodiment includes determining, based on historical data associated with a specific task, a baseline bandwidth recommendation for completing the specific task. The embodiment assigns, for a first time period, the specific task to a first computing device on a network. The embodiment allocates, for the first time period based on the baseline bandwidth recommendation, a first baseline bandwidth to the first computing device. The embodiment allocates, for the first time period, a portion of a shared buffer bandwidth as a first buffer bandwidth to the first computing device based on a weight value assigned to the specific task. The first buffer bandwidth combines with the first baseline bandwidth as a first total bandwidth for the specific task. The embodiment throttles, during the first time period, data packets associated with the specific task based on the first total bandwidth for the specific task.

Abstract: Lead to recognition time delay data, leads and sales data, and lead distance data for leads arriving at a system within a time range can be used to determine a time variable close rate and a distance adjustment thereof which, in turn, can be used to generate a predicted close rate for each lead. Using the predicted close rate for each lead, an address drive distance for each lead, and dealer-level discount information, a logistic regression model can be trained to generate a score for each combination of a dealer, make, and zip code. Each score indicates a probability of closing a sale between a lead in the zip code and the dealer in the combination. Scores thus generated can be used by a computing facility in recommending dealers to a consumer in the zip code who has expressed an interest in purchasing a vehicle of the same make.

Abstract: The present disclosure provides a computer-implemented method, computer system and computer program product for managing a task flow. According to the computer-implemented method, a definer module may receive a request for executing a task flow. The definer module may determine a cluster of edge devices to execute the task flow from a set of edge devices. The definer module may retrieve metadata information for the task flow and edge devices in the cluster, wherein the metadata information is used to schedule the task flow in the cluster. Then the edge devices in the cluster may execute the task flow according to the metadata information.

Abstract: A call processing method and call processing device are provided. The method provides a media server that receives an initial call request sent by a calling terminal and forwards the initial call request to a called terminal. The media server receives a success response to the initial call request from the called terminal, and performs video resource negotiation with the called terminal. The media server sends, based on video resource negotiation with the called terminal, a video stream to the called terminal. In an audio call process between a calling user and a called user, a video stream can be sent to the called user. This improves user experience and improves network capability utilization.

Abstract: A computer-implemented method, computer system and computer program product dynamically manage failure in an edge computing environment. According to the method, a request for executing a task may be sent to a first edge device according to a defined process, where the defined process is used to schedule tasks to be executed on edge devices. In response to the first edge device failing to execute the task, the defined process may be suspended. Then, a request for executing the task may be sent to a second edge device. A task result that is received first may be taken as the task result for the task, where the task result is from either the first edge device or the second edge device. And, continuing the rest of the defined process.

Abstract: A method for performing a directional operation includes: obtaining an operation directing to a target operation device; determining a direction of the operation, determining the target operation device matched with the direction of the operation; and performing the operation on the target operation device.

Abstract: An example fused convolutional layer, comprising, a comparator capable of reception of a first zero point and a multiply-accumulation result, a first multiplexer coupled to the comparator, wherein the first multiplexer receives a plurality of power-of-two exponent values, a shift normalizer, coupled to the first multiplexer, wherein the shift normalizer is capable of receiving the multiply-accumulation result and the plurality of power-of-two exponent values, wherein the shift normalizer limits a quantization of the multiply-accumulation result to a power-of-two scale and a second multiplexer coupled to an output of the shift normalizer, the first multiplexer and receives a second zero point and outputs an activation.

Abstract: The present disclosure discloses a method and a device for data transmission and belongs to the network communication technology. According to some embodiments, a sending end establishes a communication connection with a receiving end; the sending end splits the data packet to be transmitted into multiple data units; and the sending end transmits the multiple data units to the receiving end through multiple data streams.

Abstract: Embodiments of the present application relate to the technical field of Internet, and provide a live streaming control method and apparatus, a live streaming device, and a storage medium. Voice information of a live streamer is obtained, and the voice information is analyzed and processed, so that according to the processing result, a virtual image in a live streaming screen is controlled to execute an action matching the voice information, so as to improve the precision of controlling the virtual image and enable the virtual image in the live streaming screen and the live streaming content of the live streamer to have a high matching degree.

Abstract: A device control method, an electronic device, and a storage medium are provided. The method can include establishing a first communication connection with at least one controlled device in response to a control device entering a control state, and acquiring a preset communication index corresponding to the first communication connection. The method can further include determining relative position information with respect to the at least one controlled device according to the preset communication index, and determining one of the controlled devices as a target device based on the relative position information, and controlling the target device.

Abstract: A transportation management system determines trip defects associated with trip access points based on historical trip data. The transportation management system aggregates historical trip data in a spatial index. The transportation management system represents the spatial index using a geographic grid including grid cells at various resolutions. The transportation management system determines a defect score for individual grid cells at a given resolution based on the trip data in the spatial index corresponding to the grid cell. The transportation management system uses the defect scores for grid cells to provide a dynamic user experience to users of the transportation management system (e.g., riders or drivers) on their respective client devices.

Abstract: In one embodiment, a method includes receiving one or more querying images associated with a container of a pharmaceutical product, each of the one or more querying images being based on a particular angle of the container of the pharmaceutical product, calculating one or more confidence scores associated with one or more defect indications, respectively for the container of the pharmaceutical product, by processing the one or more querying images using a target machine-learning model, and determining a defect indication for the container of the pharmaceutical product from the one or more defect indications based on a comparison between the one or more confidence scores and one or more predefined threshold scores, respectively.

Abstract: A method includes irradiating an energy storage device using an input radiation characterized by a first electromagnetic spectrum and detecting an output radiation reflected or backscattered by the energy storage device. The method also includes determining a second electromagnetic spectrum of the output radiation and comparing the second electromagnetic spectrum with a reference electromagnetic spectrum. The method further includes generating a sorting instruction based on comparison of the second electromagnetic spectrum with the reference electromagnetic spectrum.

Abstract: A display device includes: a display unit, a first logic control unit and a light emitting driving unit. The first logic control unit is configured to receive image display data for instructing the display unit to display a static image in a local area, if it is determined that the local area maintains displaying the static image beyond a first threshold duration, then preset replacement display data and the image display data are alternately transmitted to the light-emitting driving unit, where a transmission frequency of the replacement display data is a preset maintenance frequency, and the replacement display data is used to enable the display unit to display a replacement color image in the local area; and the light-emitting driving unit is configured to respond to the data received from the first logic control unit and control the display unit to display the static image or the replacement color image.

Abstract: A robot for testing lower limb performance of a spacesuit includes a pressure maintaining box, an air circulation component, an air cooling unit, heat radiating hose components, and two mechanical legs. The air cooling unit is connected with the pressure maintaining box; the air circulation component is arranged in the pressure maintaining box; the mechanical legs are installed on the pressure maintaining box, and the heat radiating hose components are arranged in the mechanical legs; air in the pressure maintaining box is cooled through the air cooling unit and delivered into the heat radiating hose components through the air circulation component; each mechanical leg comprises a thigh, a knee joint component, a shank, an ankle joint component and a foot; the thigh is connected with the shank through the knee joint component; the shank is connected with the foot through the ankle joint component.

Abstract: A method includes processing at least one battery into a plurality of core sections. Each core section in the plurality of core sections includes an anode section, a cathode section including a cathode material, a separator section disposed between the anode section and the cathode section, and an electrolyte. The method also includes disposing the plurality of core sections into a solvent so as to produce a mixture of cathode materials from the plurality of core sections. The solvent and the electrolyte form an ionic conductive medium, and the mixture of the cathode materials is characterized by a substantially homogeneous distribution of an active element in the cathode material.

Abstract: The present disclosure relates to a terminal positioning method, apparatus, electronic device, and storage medium. The method can be applied to at least one spatial reference device in a space where a terminal is located. At least two subspaces can be included in the space where the terminal is located that are respectively provided with at least one position sensing device. Further, the spatial reference device is a device that satisfies a preset position condition among the position sensing devices. The method can include performing a first communication interaction with a terminal to be located based on a first communication protocol, determining a first relative positioning result for positioning the terminal based on a result of the first communication interaction, and determining a target subspace where the terminal is located based on the first relative positioning result.

Abstract: The present invention relates to a high-LPI lysolecithin and methods related to the same. The high-LPI product is produced through the reaction of lecithin with a unique method using solvent, buffer/water, and phospholipase. Using the current production method, LPI content increased from 1.4% to 3.2-13.1% using regular soy lecithin as the starting material and LPC, LPE and LPA contents also increased from 5.1%, 2.0%, 1.0% to 15.8%, 14.6% and 4.4% respectively.

Abstract: The application discloses a multi-type acoustic feature integration method and system based on deep neural networks. The method and system include using labeled speech data set to train and build a multi-type acoustic feature integration model based on deep neural networks, to determine or update the network parameters of the multi-type acoustic feature integration model; the method and system includes inputting the multiple types of acoustic features extracted from the testing speech into the trained multi-type acoustic feature integration model, and extracting the deep integrated feature vectors in frame level or segment level. The solution supports the integrated feature extraction for multiple types of acoustic features in different kinds of speech tasks, such as speech recognition, speech wake-up, spoken language recognition, speaker recognition, and anti-spoofing etc.