Abstract: A systolic neural CPU (SNCPU) including a two-dimensional systolic array of reconfigurable processing elements (PE's) fuses a conventional CPU with a convolutional neural network (CNN) accelerator in four phases of operation: row-CPU, column-accelerator, column-CPU, and row-accelerator. The SNCPU cycles through the four phases to avoid costly data movement across cores, reduce overhead, and reduce latency. The PE's communicate bidirectionally with neighboring PE's and memory units at an outer edge of the array. A row of PE's is configurable into a first deep neural network (DNN) accumulator at a first time and configurable into a first CPU pipeline at a second time. A column of PE's is configurable into a second DNN accumulator at a third time and configurable into a second CPU pipeline at a fourth time.

Abstract: The disclosed are shelf-stable beta-hydroxyisovaleric acid containing soft gel compositions and methods for making such compositions. In certain aspects, disclosed compositions comprise a plant-based capsule shell and a liquid formulation comprising beta-hydroxyisovaleric acid and at least one excipient, wherein the soft gel capsule is shelf-stable. In certain aspects, the excipient is selected from selected from choline salt, betaine, phosphatidylcholine, alpha-glycerophosphocholine, carnitine, adenosine 5?-triphosphate, or a combination thereof.

Abstract: The invention relates to a hybrid power drive system, comprising: an internal combustion engine having a crankshaft; a first electric motor (14), wherein the first electric motor (14) is an outer rotor electric motor, and comprises an outer rotor (14.2) that is rigidly connected to the crankshaft and rotates together with the crankshaft; a transmission (15) comprising an input shaft (20); and a clutch (18) that is provided between the first electric motor (14) and the transmission (15), and is connected to the input shaft (20) of the transmission. The clutch (18) is configured to be capable of switching between the following positions: an engagement position where the clutch (18) is engaged with the outer rotor (14.2); and a separation position where the clutch (18) is separated from the outer rotor (14.2). The present system is simple in structure, high in efficiency, and low in manufacturing and maintenance costs.

Abstract: A prediction method for water quality biotoxicity based on artificial intelligence neural network is provided, which relates to the field of water quality biotoxicity prediction technologies, and is used to solve a problem that water quality biotoxicity detection is limited to water quality organisms themselves. The prediction method uses a system including a water quality analysis unit configured to analyze simulated to-be-predicted waters to obtain variation values of water pollution, a fish analysis unit configured to analyze a fish situation of test fish to obtain fish abnormal values, a water quality determination unit configured to determine water quality of the simulated to-be-predicted waters to obtain an actual water quality level of the simulated to-be-predicted waters and send it to a data comparison unit, and the data comparison unit configured to compare data of water quality situation of the simulated to-be-predicted waters to obtain a prediction result of water quality biotoxicity.

Abstract: Implementations relate to updating agricultural records that include inferences gathered by one or more sensors deployed at an agricultural location. The records are matched to mapping tiles of a mapping application and the mapping tile records are updated according to the updated data that was received. Implementations further include identifying parent tiles to each of the updated records, where a parent tile is an aggregation of multiple mapping tiles that can be utilized by a mapping application to render an interface that allows a user to view the data with varying degrees of granularity.

Abstract: Systems and methods for a unified reconfigurable neural central processing unit is provided. In one aspect, a neural central processing unit is in communication with a memory, wherein the neural central processing unit is configured to transition between a binary neural network accelerator mode and a central processing unit mode, wherein, in the binary neural network accelerator mode, the memory is configured as an image memory and weight memories, wherein, in the central processing unit mode, the memory is reconfigured, from the image memory and the weight memories, to a data cache.

Abstract: A head-mounted device may have a head-mounted support structure, a gaze tracker in the head-mounted support structure, and one or more displays in the head-mounted support structure. For example, two displays may display images to two eye boxes. The display may display a virtual keyboard and may display a text input in response to a gaze location that is determined by the gaze tracker. The gaze tracker may additionally determine a gaze swipe input, or a camera in the support structure may determine a hand swipe input, and the swipe input may be used with the gaze location to determine the text input. In particular, the swipe input may create a swipe input curve that is fit to the text input to determine the text input. A user's hand may be used as a secondary input to indicate the start or end of a text input.

Abstract: In certain aspects, a system-on-a-chip (SoC) for human activity recognition includes a plurality of integrated low-noise amplifiers configured to sense electromyogram (EMG) signals. The SoC includes a mixed-signal circuitry configured to receive the EMG signals from the plurality of integrated low-noise amplifiers, wherein the mixed-signal circuitry is configured to digitalize and extract time-domain features from the EMG signals. The SoC includes an artificial intelligence (AI) core comprising a reconfigurable neural network (NN) configured to receive, from the mixed-signal circuitry, the time-domain features that were extracted, wherein the reconfigurable NN is configured to recognize a local gesture based on time-domain features that is extracted. The SoC includes an analog data path circuitry configured to sense distance measurements and to transmit the distance measurements and the local gesture that is recognized.

Abstract: In certain aspects, a compute-in-memory processor includes central computing units configured to operate in a central processing unit mode and a deep neural network mode. A data activation memory and a data cache output memory are in communication with the compute-in-memory processor. In the deep neural network mode, the data activation memory is configured as input memory and the data cache output memory is configured as output memory. In the central processing unit mode, the data activation memory is configured as a first data cache and the data cache output memory is configured as a register file and a second data cache.

Abstract: From an input image comprising a portion of text, the portion of text is extracted. The portion of text comprises text in a first natural language. The input image is classified into a candidate scene within a set of predefined scenes. The candidate scene is scored according to a quality measure. Using a predefined scene parameter of the scored candidate scene, the portion of text is translated. In the input image, the portion of text is replaced. with the translated portion of text.

Abstract: The disclosed are shelf-stable beta-hydroxyisovaleric acid containing soft gel compositions and methods for making such compositions. In certain aspects, disclosed compositions comprise a plant-based capsule shell and a liquid formulation comprising beta-hydroxyisovaleric acid and at least one excipient, wherein the soft gel capsule is shelf-stable. In certain aspects, the excipient is selected from selected from choline salt, betaine, phosphatidylcholine, alpha-glycerophosphocholine, carnitine, adenosine 5?-triphosphate, or a combination thereof.

Abstract: A machine learning based control scheme for a buck converter achieves fast and more energy efficient regulation of power supply change using a machine learning module that predicts upcoming supply droop or overshoot. The prediction results may be sent to a buck converter to provide power regulation in time so that the buck converter can mitigate the droop or overshoot. An event-based detection and reaction control circuit, serving as “safety net” may address mispredictions. The machine learning module may consider a microprocessor's internal operation states and earlier supply voltages. The event-based control circuit nay include both a fast droop response circuit and a slow droop response circuit that address different supply droop scenarios.

Abstract: Aspects of the present disclosure relate generally to UI automation in RPA and, more particularly, to automation of contextual UI element identification by RPA robots. For example, a computer-implemented method includes: receiving, by a processor set, user interface context identification information of attributes of user interface elements in user interface code of an application; constructing, by the processor set, a model of the attributes of the user interface elements including an attribute indicating functionality of at least one user interface element; obtaining from the model of the attributes of the user interface elements an identification of the at least one user interface element referenced in an action command for performing the functionality of the at least one user interface element in robotic process automation code; and storing, by the processor set, the model of the attributes of the user interface elements in persistent storage.

Abstract: The present disclosure relates to the technical field of pyrolysis and improvement of caking middling coals, in particular to a low temperature pyrolysis method of a caking middling coal. The present disclosure provides a low temperature pyrolysis method of a caking middling coal, including the following steps: conveying the caking middling coal into a pyrolysis reactor through a top of the pyrolysis reactor; dividing a reaction chamber of the pyrolysis reactor into a drying section, a softening section, a melting and depolymerization section, a solidification section, and a cooling section by means of multi-channel gas distribution; and conducting zoned temperature control-based pyrolysis to obtain semi-coke at a bottom of the reactor as well as tar and coal gas at the top of the reactor. The pyrolysis method can well avoid caking and swelling of the caking middling coal during pyrolysis.

Abstract: Systems formed by a multi-bit three-transistor (3T) memory cell (i.e., dynamic-analog RAM) are provided. The 3T memory cell includes: a read-access transistor M1 in electrical communication with a read bitline; a switch transistor M2 in electrical communication with the read-access transistor M1; a write-access transistor M3 in electrical communication with the read-access transistor M1 and a write bitline; and a memory node MEM in electrical communication between the read-access transistor M1 and the write-access transistor M3, wherein the memory node MEM is configured to store a 4-bit weight WE. An array of the 3T memory cells (i.e., dynamic-analog RAMs) may form a computing-in-memory (CIM) macro, and further form a convolutional neural network (CNN) accelerator by communicating with an application-specific integrated circuit (ASIC) which communicates with a global weight static random access memory and an activation static random access memory.

Abstract: Implementations relate to updating agricultural records that include inferences gathered by one or more sensors deployed at an agricultural location. The records are matched to mapping tiles of a mapping application and the mapping tile records are updated according to the updated data that was received. Implementations further include identifying parent tiles to each of the updated records, where a parent tile is an aggregation of multiple mapping tiles that can be utilized by a mapping application to render an interface that allows a user to view the data with varying degrees of granularity.

Abstract: The present disclosure relates to the technical field of pyrolysis and improvement of caking middling coals, in particular to a low temperature pyrolysis method of a caking middling coal. The present disclosure provides a low temperature pyrolysis method of a caking middling coal, including the following steps: conveying the caking middling coal into a pyrolysis reactor through a top of the pyrolysis reactor; dividing a reaction chamber of the pyrolysis reactor into a drying section, a softening section, a melting and depolymerization section, a solidification section, and a cooling section by means of multi-channel gas distribution; and conducting zoned temperature control-based pyrolysis to obtain semi-coke at a bottom of the reactor as well as tar and coal gas at the top of the reactor. The pyrolysis method can well avoid caking and swelling of the caking middling coal during pyrolysis.

Abstract: Adaptively controlling a speaker assembly allows an acoustic characteristic of the speaker assembly to automatically change based on one or more factors. For example, when external noise is detected, the associated noise volume level is determined, and control instructions are generated based on the noise volume level. The control instructions cause the speaker assembly to adaptively increase the volume level to effectively cancel the noise volume. In another example, when a speaker assembly in a vehicle changes from playing a media file to conducting a phone conversation, the volume level associated with the phone conversation may increase. In this regard, control instructions can be generated and used to adaptively reduce the volume level of the phone conversation.

Abstract: Disclosed is a stator structure of a flat wire motor, including a stator core, a first conductor, a second conductor and a third conductor. An inner wall of the stator core is provided circumferentially and uniformly with m stator slots; the first conductor, the second conductor and the third conductor are provided in the stator slots. The present application uses a 2-way N-layer winding solution to provide a flat wire winding connection solution with balanced three-phase resistance and no loop current by connecting six neutral point conductors, a first conductor, six second conductors and six third conductors; the three-phase outlet wire conductors and neutral point copper row are in two adjacent layers, and positions are concentrated after twisting the head to simplify the end structure.

Abstract: Disclosed are a class of URAT1 inhibitor compounds and the use of such compounds. These compounds are compounds represented by the structure of formula (I) or pharmaceutically acceptable salts thereof. Experiments show that the compounds provided by the present invention have a very good inhibitory effect on hURAT1-transported uric acid in HEK293 transfected cells, and show that such compounds have a good potential for application in the treatment of hyperuricemia or gout.