Abstract: A neural network compression method whereby forward inference is performed on target data by using a target parameter sharing network to obtain an output feature map of the last convolutional module, a channel related feature is extracted from the output feature map, the extracted channel related feature and a target constraint condition are input into a target meta-generative network, and an optimal network architecture under the target constraint condition is predicted by using the target meta-generative network to obtain a compressed neural network model.

Abstract: A connector for use with high-speed signals. The connector includes conductive elements held in rows and a member between rows of conductive elements. The conductive elements have pairs of signal conductors and ground conductors disposed between the pairs of the signal conductors. The member has a lossy body and a conductive layer plated on the lossy body to electrically connect the ground conductors. The lossy body of the member has openings and/or protrusions disposed and sized such that the conductive layer plated on the lossy body provides shortened conductive paths between ground conductors in different rows. The member can be configured to tune the impedance at desired regions such as the mating region. Such a configuration meets signal integrity requirements in connectors designed for high speed transmission, while conforming to a standard that constrains mating and mounting interfaces.

Abstract: A hardware environment-based data operation method, apparatus and device, and a storage medium. The method includes: determining data to be operated and target hardware, wherein the target hardware is a hardware resource that needs to perform convolution computation on the data to be operated currently; determining the maximum number of channels in which the target hardware executes parallel computation, and determining a data layout corresponding to the maximum number of channels to be an optimal data layout; and converting the data layout of the data to be operated into the optimal data layout, and performing the convolution computation on the data to be operated by using the target hardware after the conversion is completed.

Abstract: An image processing method, apparatus and device, and a readable storage medium are disclosed, including: obtaining a target image; inputting the target image into a quantized target deep neural network model for classification/detection to obtain an output result; and processing the target image according to a policy corresponding to the output result. A process of performing quantization to obtain the target deep neural network model includes: obtaining a pre-trained floating point type deep neural network model; extracting weight features of a deep neural network model; determining a quantization policy using the weight features; and quantizing the deep neural network model according to the quantization policy to obtain the target deep neural network model.

Abstract: Plug connectors for use with high speed signals. The connector includes an insulative housing having a base portion and a tongue portion extending from the base portion, and conductive terminals held by the housing. Each conductive terminal has a mating portion held on a side of the tongue portion, a tail portion extending out of the housing, and an intermediate portion held in the base portion. The conductive terminals comprise signal terminals and ground terminals. A shielding member is disposed in the housing and includes features configured to provide shielding to signal terminals and electrically connect ground terminals. The shielding member also includes features configured to improve impedance consistency along conduction paths. Such a configuration meets signal integrity requirements in connectors designed for higher frequencies, while conforming to a standard that constrains mating and mounting interfaces.

Abstract: An image recognition method and device, and a computer-readable storage medium. The method includes: inputting a sample image data set into an original neural network model in advance; and for each convolution layer of the original neural network model, using a feature map of the sample image data set in the current layer as a reconstruction target, firstly obtaining an update weight of a convolution kernel by using a kernel set construction method, then calculating an input channel combination having a minimum reconstruction error, cutting a redundant input channel to obtain a compression result of the current convolution layer, and finally, splicing compression results of the convolution layers to generate an image recognition model.

Abstract: A neural network compression method, apparatus and device, and a storage medium are provided. The method includes: performing forward inference on target data by using a target parameter sharing network to obtain an output feature map of the last convolutional module; extracting a channel related feature from the output feature map; inputting the extracted channel related feature and a target constraint condition into a target meta-generative network; and predicting an optimal network architecture under the target constraint condition by using the target meta-generative network to obtain a compressed neural network model. By using the technical solution, the computation load of a performance evaluation process of a neural architecture search may be reduced, and the speed of the searching for a high-performance neural network architecture may be increased.

Abstract: A hardware environment-based data quantization method includes: parsing a model file under a current deep learning framework to obtain intermediate computational graph data and weight data that are independent of a hardware environment; performing calculation on image data in an input data set through a process indicated by an intermediate computational graph to obtain feature map data; separately performing uniform quantization on the weight data and the feature map data of each layer according to a preset linear quantization method, and calculating a weight quantization factor and a feature map quantization factor (S103); combining the weight quantization factor and the feature map quantization factor to obtain a quantization parameter that makes hardware use shift instead of division; and finally, writing the quantization parameter and the quantized weight data to a bin file according to a hardware requirement so as to generate quantized file data (S105).

Abstract: A connector for use with high speed signals. The connector has a housing, which includes a base portion extending in a longitudinal direction and a tongue portion extending from the base portion in a mating direction perpendicular to the longitudinal direction. The housing has slots extending through the base portion to surfaces of the tongue portion and configured to hold terminals. The housing has a lossy material sized and shaped to couple with at least the portions of ground terminals held in the base portion along their lengths. The lossy material may protrude towards spaces between the portions of signal terminals held in the base portion along their lengths. The tail portions of the terminals may jog toward a plane that the lossy material extends. Such a connector may be used to meet signal integrity requirements in connectors designed for 64 Gbps and beyond.

Abstract: A hardware environment-based data quantization method includes: parsing a model file under a current deep learning framework to obtain intermediate computational graph data and weight data that are independent of a hardware environment; performing calculation on image data in an input data set through a process indicated by an intermediate computational graph to obtain feature map data; separately performing uniform quantization on the weight data and the feature map data of each layer according to a preset linear quantization method, and calculating a weight quantization factor and a feature map quantization factor (S103); combining the weight quantization factor and the feature map quantization factor to obtain a quantization parameter that makes hardware use shift instead of division; and finally, writing the quantization parameter and the quantized weight data to a bin file according to a hardware requirement so as to generate quantized file data (S105).

Abstract: A connector for use with high-speed signals. The connector may include a row of conductive elements comprising signal conductors and ground conductors. A signal conductor may include a mating end, a mounting end opposite the mating end, and an intermediate portion extending therebetween. A ground conductor may include a shell at least partially encircling intermediate portions of a group of signal conductors, with openings that expose contact surfaces of the signal conductors at the ends. The ground conductors may include contact members aligned with the mating contact surfaces of the signal conductors. Each shell may be coupled to a ground mounting location on each side of a respective group of signal conductors, with adjacent shells sharing a ground mounting location. Such a configuration may meet signal integrity requirements in connectors designed for 64 Gbps and beyond, while conforming to a standard that constrains mating and mounting interfaces.

Abstract: A safety drilling system for preventing collapse of water-sensitive formation in the upper part of a high-pressure saltwater layer, which includes a wellhead equipment, a downhole drilling tool, a first injection pipeline, a second injection pipeline, a first return pipeline and a second return pipeline. Upon encountering a high-pressure saltwater layer while drilling, the system injects heavy mud into the annulus of a wellhead through the first injection pipeline to form a heavy mud cap, such that a fluid column pressure in the annulus balances the pressure of the high-pressure saltwater layer. In this case, a fluid column pressure in a drill string is lower than the pressure of the high-pressure saltwater layer.

Abstract: The invention introduces a downhole failure analysis and processing method and device based on the partical diameter distribution of drilling cuttings. The device includes body frame, screening component, feeding system, weighing mechanism, driving mechanism and control system, which can realize automatic grading and screening according to the partical diameter of cuttings and weigh the cuttings at all levels; the method adopts the debris sieving device to measure the particle diameter distribution of cuttings, establishes a standard partical diameter distribution database of cuttings, compares with the real-time partical diameter distribution of cuttings to detect the downhole failure, and then selects the corresponding failure solution from the downhole failure data to remove the downhole failure.

Abstract: A safety drilling system for preventing collapse of water-sensitive formation in the upper part of a high-pressure saltwater layer includes a wellhead equipment, a downhole drilling tool, a first injection pipeline, a second injection pipeline, a first return pipeline and a second return pipeline, and further includes a drilling method, wherein after encountering a high-pressure saltwater layer while drilling, heavy mud is injected into the annulus of a wellhead through the first injection pipeline to form a heavy mud cap, such that a fluid column pressure in the annulus balances the pressure of the high-pressure saltwater layer; in this case, a fluid column pressure in a drill string is lower than the pressure of the high-pressure saltwater layer.

Abstract: A method for establishing a chart for designing mechanical properties of cement stones in a large-scale fracturing oil well is provided, including steps of: establishing a stress distribution model of a cement sheath based on a theory of elasticity and thick-walled cylinder; establishing a cement sheath integrity prediction model based on the cement sheath failure criterion and the stress increment distribution state of the cement sheath; establishing a cement sheath integrity control method based on the cement sheath stress analysis model and the cement sheath integrity prediction model; establishing a functional relationship between cement stone mechanical parameters and strength parameters based on the cement sheath integrity control method; and establishing a cement stone performance index control chart based on the functional relationship between the cement stone mechanical parameters and strength parameters.

Abstract: The invention introduces a downhole failure analysis and processing method and device based on the partical diameter distribution of drilling cuttings. The device includes body frame, screening component, feeding system, weighing mechanism, driving mechanism and control system, which can realize automatic grading and screening according to the partical diameter of cuttings and weigh the cuttings at all levels; the method adopts the debris sieving device to measure the particle diameter distribution of cuttings, establishes a standard partical diameter distribution database of cuttings, compares with the real-time partical diameter distribution of cuttings to detect the downhole failure, and then selects the corresponding failure solution from the downhole failure data to remove the downhole failure.