Abstract: Methods and systems for training a neural network include collecting model exemplar information from edge devices, each model exemplar having been trained using information local to the respective edge devices. The collected model exemplar information is aggregated together using federated averaging. Global model exemplars are trained using federated constrained clustering. The trained global exemplars are transmitted to respective edge devices.

Abstract: A curable composition that includes a urethane multifunctional (meth)acrylate, an inorganic filler having a primary particle dimension of at least 200 nm, a photoinitiator system that can be activated by electromagnetic radiation in the range of 340-550 nm, a reactive diluent, and a reinforcing silica having a primary particle dimension of 100 nm of less. The sum of the absolute value of the difference in the refractive index of the filler and the refractive index of the composition cured without filler plus the birefringence of the filler is 0.054 or less, i.e. 0.054?|nfiller?nmatrix|+?filler, where nfiller is the refractive index of the filler, nmatrix is the refractive index of the composition cured without filler, and ?filler is the birefringence of the filler.

Abstract: A temperature control method for a vehicular proton exchange membrane fuel cell system comprises the following steps: detecting a cooling loop inlet temperature of a fuel cell stack by using a temperature sensor, and inputting the temperature into a controller to achieve cooling fan control based on the controller, wherein the cooling fan control comprises fuzzy logic self-adaptive proportional integral control and feedforward compensation control, gain parameters of the proportional integral control are self-adaptively updated by a fuzzy logic algorithm, a load current of the fuel cell serves as disturbance and is used for feedforward compensation, and meanwhile, the opening degree of the fan is determined by the total cooling capacity requirement and the number of cooling fans; and finally, inputting a control signal output by the controller into an actuator of a thermal management subsystem, and conducting cooling inlet temperature control of the fuel cell stack.

Abstract: A piezoelectric microphone, includes: a wafer substrate including a cavity; a plurality of cantilever beams with a piezoelectric deck structure; a fixed column; a plurality of flexible elastic members; and a connecting section. The plurality of cantilever beams each includes a fixed end and a free end suspended above the cavity. The plurality of cantilever beams is of a structure in which one end is narrow and the other end is wide, and the fixed end is relatively narrow. The fixed column is disposed at the center of the bottom surface of the cavity. The fixed ends of the plurality of cantilever beams are all connected to the top surface of the fixed column. A gap is provided between every two adjacent cantilever beams. The plurality of flexible elastic members is connected to free ends of two adjacent cantilever beams to enable the cantilever beams to vibrate synchronously.

Abstract: Provided are a chimeric antigen receptor targeting CD20 antigen and a preparation method thereof. The extracellular antigen binding domain of the chimeric antigen receptor includes an antibody heavy chain variable region shown in SEQ ID NO: 7 or 9 or 33 and an antibody light chain variable region shown in SEQ ID NO: 11 or 13 or 35, and is capable of killing tumor cells.

Abstract: A method to prepare a population of metal-organic polyhedra (MOP) supported micelle nanoparticles (NPs), and a composition comprising MOP supported micelle NPs, are provided.

Abstract: Dew point sensor system contains-a controller with dew point calculator, ambient temperature sensor, air pathway, heater plate, external wire. Ambient temperature sensor generates ambient temperature data, operatively-connected to controller. Air pathway contains blower within air pathway, air flow sensor within air pathway, liquid reservoir operatively-connected to air pathway, and component may be air pathway temperature sensor and/or breathing circuit heating element. Blower causes air to move in air pathway (52), air flow sensor generates air flow data about air moving in air pathway. The heater plate contains heater plate temperature sensor, and heater plate temperature sensor generates heater plate temperature data. External wire is operatively-connected to controller and component. Ambient temperature sensor located on the external wire.

Abstract: A quantum dot film, a quantum dot light-emitting assembly and a display device are provided. The quantum dot film includes: a quantum dot layer; and a conductive layer arranged on at least a side of the quantum dot layer along a thickness direction, and the conductive layer includes nano-sized metal particles, and at least a portion of the nano-sized metal particles are configured to generate a surface plasmon resonance under electromagnetic radiation. The luminescence efficiency and intensity of the quantum dot layer can be effectively improved by arranging the conductive layer on at least a side of the quantum dot layer.

Abstract: The present invention relates to compounds of formula (I), wherein R1, R2, R3, R4 and X are as described herein, and their pharmaceutically acceptable salt, enantiomer or diastereomer thereof, and compositions including the compounds and methods of using the compounds.

Abstract: The present invention discloses a method of hot recycling repairing by optimizing a proportion of an asphalt mixture on a pavement alignment variation section. According to the method of the present invention, by regulating and optimizing the proportion of a recycled asphalt mixture, internal frictional resistance and cohesion of the recycled asphalt mixture are improved, elastic modulus and shear strength of the recycled asphalt mixture are increased, overall bearing capacity of an asphalt pavement is improved, and diseases such as tracks caused by vehicle loads are effectively resisted. The pavement asphalt mixture is made to adapt to and be enough to bear additional loads increased manifold without generating early defects and road diseases. Therefore, the overall service life of a road is further prolonged.

Abstract: A combined remediation method of petroleum-contaminated soil includes: impurity removal pretreatment, photocatalytic pre-oxidation, stepwise thermal desorption of petroleum from soil, and high-temperature oxidation; with iron-titanium composite metal oxide (ITCMO) as a catalyst, conducting oxidation pretreatment under light conditions so that some cross-linked structures in macromolecular petroleum contaminants are broken and degraded; and conducting stepwise pyrolysis to achieve a removal rate of more than 98.00%. The new method adopts a combined remediation technology of photocatalytic pre-oxidation-stepwise pyrolysis, which realizes a relatively-high removal rate of petroleum hydrocarbons and the efficient and harmless remediation of high-concentration petroleum-contaminated soil, and remedied soil can be reused.

Abstract: A laser printing device for making high-precision 3D products includes a platform, a device supplying and spreading powder for transporting and spreading metal powder on the top surface of the platform, a first laser source for emitting continuous laser beam, a second laser source for emitting pulsed laser beam, an optical path switching device connected with the first laser source and the second laser source for controlling and selecting the first laser source and the second laser source to emit laser beam, a scanning galvanometer system for focusing the laser beam on the platform for making the continuous laser beam melt the metal powder and a layer of a product being formed after cooling, a mechanical processing device for polishing the final surface of the product. The pulsed laser beam cuts the edge of each newly-deposited layer with geometric accuracy. A laser printing method for 3D product is also provided.

Abstract: The disclosure provides an intelligent processing apparatus including a processor and a direct memory access (DMA) controller. The processor generates a read command to read data from a memory. An address generation circuit generates multiple first address signals to the memory. A data processing circuit receives data that the memory outputs in response to the multiple first address signals, and performs a data interception process on the data received from the memory according to the read command and a first configuration parameter. A data port transmits the data processed by the data processing circuit to the processor. The multiple first address signals include multiple non-consecutive address signals such that the data read from the memory and processed by the data processing circuit corresponds to n-dimensional data blocks, where n is a positive integer greater than 1.

Abstract: A convolution calculation apparatus applied for convolution calculation of a convolution layer includes a decompression circuit, a data combination circuit and a calculation circuit. The decompression circuit decompresses compressed weighting data of a convolution kernel of the convolution layer to generate decompressed weighting data. The data combination circuit combines the decompressed weighting data and non-compressed data of the convolution kernel to restore a data order of weighting data of the convolution kernel. The calculation circuit performs calculation according to the weighting data of the convolution kernel and input data of the convolution layer.

Abstract: A convolutional neural network operation device includes a scheduling mode unit, a first data processing circuit, a second data processing circuit and a multiple-accumulate (MAC) operation array. The scheduling mode unit determines, according to a quantity and size information of the target convolutional kernels, a target scheduling mode corresponding to a size of a convolutional computing block. The first data processing circuit recombines weight data in the target convolutional kernels and the second data processing circuit recombines input data in a target convolutional layer according to the target scheduling mode. The MAC operation array includes multiple MAC operation cells, and performs a MAC operation based on the recombined weight data and the recombined input data, wherein a quantity of the MAC operation cells used by the MAC operation array in each round of operation corresponds to the size of the convolutional computing block.

Abstract: The present invention relates to a pyrimidine and a five-membered nitrogen heterocycle derivative, a preparation method therefor, and the medical uses thereof. Particularly, the present invention relates to a pyrimidine and a five-membered nitrogen heterocycle derivative represented by the general formula (I), a preparation method thereof, a pharmaceutical composition containing the derivative, and the uses thereof as a SHP2 inhibitor for use in the prevention and/or treatment of tumor or cancer, wherein each substituent in the general formula (I) is as defined in the description.

Abstract: A virtual memory management method applied to an intelligent processor including an operation accelerator includes: determining m storage units from a physical memory, the m storage units forming a virtual memory; dividing the m storage units into n storage groups; determining an address mapping relationship for each storage group to obtain n address mapping relationships, the n address mapping relationship being correspondence of between n virtual addresses of the virtual memory and physical addresses of the m storage units, where m and n are dynamically updated according to requirements of the operation accelerator. In the method, the number of the storage units in each storage group can be configured according to requirements of the operation accelerator, and a data storage bit width and a data storage depth of the virtual memory are dynamically updated to thereby improve data access efficiency.

Abstract: A transdermal peptide with a nuclear localization ability and having an amino acid sequence as shown in SEQ ID NO: 1 is disclosed. A fusion protein including a macromolecular protein with one end being linked to the transdermal peptide is also disclosed. The transdermal peptide can be used in the preparation of a medicament or a transdermal preparation for treating skin diseases. A medicament for treating a skin disease includes the transdermal peptide and a pharmaceutically acceptable excipient. The transdermal peptide enters the cells autonomously to locate in the nuclei, and can penetrate through the stratum corneum of the skin into cells in the dermis. The peptide is conveniently synthesized artificially and suitable for transdermal administration, and has a therapeutic potential via transdermal administration by carrying a drug for treating skin diseases.

Abstract: T cell receptors that specifically recognize hAFP158 and methods of their use are provided.

Abstract: Methods and systems for training a neural network include collecting model exemplar information from edge devices, each model exemplar having been trained using information local to the respective edge devices. The collected model exemplar information is aggregated together using federated averaging. Global model exemplars are trained using federated constrained clustering. The trained global exemplars are transmitted to respective edge devices.