Abstract: A personal identification apparatus includes a recording unit that records in advance feature information concerning an activity of at least one of a first person (person A) and a second person (person B or C) when the first person and the second person are together; an information acquiring unit that acquires identification information for identifying the other person; and an identification processing unit that identifies the other person as the second person, based on matching between the identification information and the feature information.

Abstract: In a method for improving converter and compiler operator granularity, a processor extracts an operator granularity from an artificial intelligence framework and an original model. A processor receives device characteristics from a target device. A processor outputs a converter granularity level to a converter based on the operator granularity and the device characteristics. A processor outputs a compiler granularity level to a compiler based on the operator granularity and the device characteristics.

Abstract: A method for quantizing a machine learning model during an inference phase, including determining a normalization factor using a set of floating-point values and a damped value of a damped value sequence; and assigning a quantized value for each floating-point value of the set of floating-point values based on the damped value sequence and the normalization factor.

Abstract: Methods and devices, the method including receiving a matrix of a neural network model; classifying at least a portion of the matrix as a first section based on a first distribution pattern of non-zero elements of the portion of the matrix; and identifying memory addresses of the non-zero elements in the first section of the matrix for loading, according to a first order determined based on the first distribution pattern, the non-zero elements in the first section into one or more vector registers.

Abstract: Aspects of the invention include converting an artificial intelligence (AI) model generated in a first framework to a uniform exchange formatted model by engaging a master table to retrieve instructions for converting from the AI model to the uniform exchange formatted model in accordance with the first framework. The uniform exchange formatted model in compiled by engaging the master table to retrieve instructions for compiling the uniform exchange formatted model in accordance with the first framework. Data is received as an input to the compiled uniform exchange formatted model and an output is generated by engaging the master table to retrieve instructions for generating the output in accordance with the first framework.

Abstract: A computer-implemented method includes assigning an auditor weight to each user of a plurality of users. A plurality of audit data is received from the plurality of users, where each audit data is associated with a respective user and includes a set of user ratings for an audit. Final ratings are generated for the audit, with a respective final rating for each item in the audit. Generating the final rating of a first item in the audit includes identifying, in the audit data, user ratings for the first item. The user ratings for the first item include a respective user rating of the first item from each user. Generating the final rating of a first item further includes calculating the final rating of the first item based on a weighted combination of the plurality of user ratings for the first item, according to the respective auditor weight of each user.

Abstract: The present disclosure relates to the technical field of solid waste recycling and fabricated buildings, and provides a solid waste large-mixing-amount concrete prefabricated laminated slab and a preparation method thereof. The solid waste large-mixing-amount concrete prefabricated laminated slab provided by the present disclosure comprises a prefabricated layer and a laminated layer. Transverse grooves and longitudinal grooves are formed in the surface of the prefabricated layer. During application, the grooves can be used for erecting pipelines, the contact area of the prefabricated layer and the laminated layer can also be increased, the combined effect of new concrete and old concrete is improved, the integrity of a floor slab is enhanced, and the effect of improving the overall stress capacity of the floor slab is achieved.

Abstract: Methods and devices, the method including receiving a matrix of a neural network model; classifying at least a portion of the matrix as a first section based on a first distribution pattern of non-zero elements of the portion of the matrix; and identifying memory addresses of the non-zero elements in the first section of the matrix for loading, according to a first order determined based on the first distribution pattern, the non-zero elements in the first section into one or more vector registers.

Abstract: Multi-solid waste activated concretes with high-silicon iron ore tailings and preparation methods thereof are disclosed. In at least some embodiments, the concrete is prepared from raw materials including 360-380 kg/m3 of a cement, 30-40 kg/m3 of fly ash, 30-40 kg/m3 of a modified ultrafine sand of high-silicon iron ore tailings, 930-950 kg/m3 of a waste stone of tailings, 870-930 kg/m3 of a fine sand of tailings, 160-170 kg/m3 of water, and 4-8 kg/m3 of an additive.

Abstract: A superhydrophobic coating is provided and contains organosilane, an inorganic nanomaterial, and an emulsifying agent. A mass proportion of the components is controlled, so that the superhydrophobic coating can form a micro-nano mixed microstructure inside foam concrete. The organosilane first forms dense hydrophobic surface layers on the surface and in inner pores of the foam concrete, and the nanomaterial forms uniformly distributed nano-bulges on the hydrophobic surface layers formed by the silane. The superhydrophobic performance of the foam concrete can be effectively improved by combining the two microstructures. The foam concrete exhibits excellent superhydrophobic performance.

Abstract: A method of manufacturing a press-fit apparatus suitable for press-fitting multiple connectors to a circuit board is provided. A machine press head that is movable by a power supplier of the press-fit apparatus is provided. A universal press block is fixed to the machine press head to allow the universal press block to be movable with the machine press head. A locating block is fixed to the bottom plate of the universal press block, by inserting a locating fastener of the locating block into a locating hole. A press block is fixed to the locating block at a predetermined position suitable for press-fitting one of the connectors.

Abstract: A method of forming solder paste on an object is provided. The method includes providing a multilayered stencil, each stencil layer having an aperture formed therethrough from a top surface of the respective stencil layer to a bottom surface of the stencil layer. The method also includes applying the stencil to a surface of the object. The method also includes filling a void space formed by a combination of the apertures in the stencil layers with solder paste. The method also includes removing the stencil layers in a sequential manner to leave the solder paste.

Abstract: Embodiments of the disclosure provide systems and methods for allocating memory space in a memory device. The system can include: a memory device for providing the memory space; and a compiler component configured for: receiving a request for allocating a data array having a plurality of data elements in the memory device, wherein each of the plurality of data elements has a logical address; generating an instruction for allocating memory space for the data array in the memory device based on the request; generating device addresses for the plurality of data elements in the memory device based on logical addresses of the plurality of data elements; and allocating the memory space for the data array in the memory device based on the device addresses and the instruction.

Abstract: Certain aspects of the present disclosure provide techniques for a feature management platform to asynchronously implement an ensemble of machine learning models (or AI/ML models). The feature management platform can transmit feature data presently available on the feature queue (e.g., published on the feature queue or retrieved from a persistent data layer). The feature management platform can transmit the feature data to a first group of the machine learning models, capable of inputting the presently available feature data to generate predictions. The predictions can be transmitted back to the feature queue to consume—as well as store in the persistent data layer. The predictions, as well as any newly generated feature data, can be provided to the remaining machine learning models in the ensemble. The prediction of the ensemble can then provided to a consumer (e.g., an organization).

Abstract: A variable-speed magnetic coupling having a radially movable magnet, comprising a drive disc assembly (I), a driven disc assembly (II), and a speed adjusting device assembly (III). Relative rotation of a speed adjustment sleeve (15) with respect to a drive shaft (16) is achieved by means of contact and fitting of a cylindrical pin (20) with respect to a vertical recess on an inner wall of the drive shaft and to an inclined recess on an inner wall of the speed adjustment sleeve. The speed adjustment sleeve is connected to a circular slotted disc (18) by means of a screw. A permanent magnet (10) is attached to a permanent magnet bearer (9) and inserted into a rectangular through hole of a circular frame (13), and a radial movement of the permanent magnet is enabled by means of a cam and groove sliding block mechanism consisting of the circular slotted disc, the circular frame, the permanent magnet bearer, and the permanent magnet.

Abstract: An operation control device includes a line-of-sight detector configured to detect a line of sight of a driver of a vehicle on the basis of an image of an interior of the vehicle captured by an imaging device provided in the vehicle, a hand state detector configured to detect a hand state of the driver on the basis of the image, a driving state determiner configured to determine whether the detected hand state is a first state in which it is estimated that a driving operation of the vehicle is being performed or a second state in which it is estimated that the driving operation is not being performed, a traveling state detector configured to detect a traveling state of the vehicle, and an operation restrictor configured to restrict reception of an operation on an operator including a display device arranged within the vehicle.

Abstract: Embodiments of the disclosure provide systems and methods for allocating memory space in a memory device. The system can include: a memory device for providing the memory space; and a compiler component configured for: receiving a request for allocating a data array having a plurality of data elements in the memory device, wherein each of the plurality of data elements has a logical address; generating an instruction for allocating memory space for the data array in the memory device based on the request; generating device addresses for the plurality of data elements in the memory device based on logical addresses of the plurality of data elements; and allocating the memory space for the data array in the memory device based on the device addresses and the instruction.

Abstract: Embodiments of the present disclosure introduce an offset-based representation for a computation graph of a neural network model. In this representation, the edges in the graph are implemented as offsets to the base address of the graph when the model is loaded into memory rather than pointers to address locations in memory. By replacing the pointer fields with offsets, the model in memory is no longer dependent on the allocation of memory. This overcomes the limitations of performance and/or power that is critical for comparing two models and copying a model to another memory location in latency sensitive applications and in mobile environments.

Abstract: Methods and devices, the method including receiving a matrix of a neural network model; classifying at least a portion of the matrix as a first section based on a first distribution pattern of non-zero elements of the portion of the matrix; and identifying memory addresses of the non-zero elements in the first section of the matrix for loading, according to a first order determined based on the first distribution pattern, the non-zero elements in the first section into one or more vector registers.

Abstract: The present disclosure relates to a method for detecting error in executing a computation graph on heterogeneous computing devices. The method comprises receiving a first reference value as an execution result for a first node of the computation graph from a reference device included in the heterogeneous computing devices, receiving a first target value from a target device included in the heterogeneous computing devices as an execution result by the target device for the first node, comparing the first reference value and the first target value, and determining whether the first target value is in error based on the comparison of the first reference value and the first target value. The method can further comprise generating multiple execution contexts for executing the computation graph on the heterogeneous computing devices.