Abstract: The present disclosure discloses a method for quantifying a bearing capacity of foundation containing shallow-hidden spherical cavities, comprising: in Step 1, constructing a spatial axisymmetric calculation model for stability analysis of the foundation containing shallow-hidden spherical cavities; in Step 2, solving the model to obtain a general solution which reflects the spatial stress distribution of surrounding rock containing shallow-hidden spherical cavities; in Step 3, obtain a mathematical expression by derivation for calculating the bearing capacity of the foundation containing shallow-hidden spherical cavities; and in Step 4: completing the determination of the foundation bearing capacity. Benefits: This method has many advantages such as comprehensive consideration, high accuracy and reliability of calculation results, and may provide the scientific basis for the development of prevention and control against the instability of the foundation containing shallow-hidden cavities.

Abstract: Receiving and storage of energy, and particularly a particle solar receiver for solar thermal power generation. The particle solar receiver includes: a feeding bin temporarily storing endothermic particles to be heated, and a multi-stage plate heat absorbing channel allowing the particles to flow along a predetermined path by gravity, where the multi-stage plate heat absorbing channel includes a plurality of plate-type structures which forms a changing flow direction of the particles flowing between adjacent plate-type structures. The particle solar receiver can achieve the endothermic particle heating function with a simple structure, i.e., the function of converting the solar power into thermal energy at a high-temperature level.

Abstract: In certain aspects, a three-dimensional (3D) memory device includes channel structures in a first region, word line pick-up structures in a dielectric portion of a second region, and word lines each extending in the first region and a conductive portion of the second region. The first region and the second region are arranged in a first direction. The dielectric portion and the conductive portion of the second region are arranged in a second direction perpendicular to the first direction. The word lines are discontinuous in the dielectric portion of the second region and are electrically connected to the word line pick-up structures, respectively.

Abstract: In certain aspects, a method for forming a three-dimensional (3D) memory device is disclosed. A stack structure including interleaved first dielectric layers and second dielectric layers is formed. Channel structures extending through the first dielectric layers and the second dielectric layers in a first region of the stack structure are formed. All the second dielectric layers in the first region and parts of the second dielectric layers in a second region of the stack structure are replaced with conductive layers. Word line pick-up structures extending through the first dielectric layers and remainders of the second dielectric layers in the second region of the stack structure are formed at different depths, such that the word line pick-up structures are electrically connected to the conductive layers, respectively, in the second region of the stack structure.

Abstract: A CO2 gas-liquid phase transition-based multistage compression energy storage apparatus for converting thermal energy into mechanical energy, including: a gas storage; a liquid storage tank; an energy storage assembly, which includes compressors and energy storage heat exchangers; an energy release assembly (400), which includes energy release heat exchangers and expanders; a heat exchange assembly the energy generated by the energy storage assembly, and the energy release heat exchangers being capable of receiving the energy temporarily stored by the heat exchange assembly; and a driving assembly, which includes an energy input member and a first driving member, the energy input member absorbing external thermal energy to drive the first driving member to work, and the first driving member being used for driving the compressors to work.

Abstract: A multi-resistance-state spintronic device, including: a top electrode and a bottom electrode respectively connected to a read-write circuit; and a magnetic tunnel junction between two electrodes. The magnetic tunnel junction includes from top to bottom: a ferromagnetic reference layer, a barrier tunneling layer, a ferromagnetic free layer, and a spin-orbit coupling layer. Nucleation centers are provided at two ends of the ferromagnetic free layer to generate a magnetic domain wall; the spin-orbit coupling layer is connected to the bottom electrode, and when a write pulse is applied, an electron spin current is generated and drives the magnetic domain wall through a spin-orbit torque to move; a plurality of local magnetic domain wall pinning centers are provided at an interface between the spin-orbit coupling layer and the ferromagnetic free layer to enhance a strength of a DM interaction constant between interfaces.

Abstract: The three-state spintronic device includes: a bottom electrode, a magnetic tunnel junction and a top electrode from bottom to top. The magnetic tunnel junction includes: a spin-orbit coupling layer, a ferromagnetic free layer, a barrier tunneling layer, a ferromagnetic reference layer, three local magnetic domain wall pinning centers and domain wall nucleation centers. An antisymmetric exchange interaction is modulated, and the magnetic domain wall pinning centers are embedded in an interface between a heavy metal and the ferromagnetic free layer. The magnetic domain wall nucleation centers are at two ends of the ferromagnetic free layer. A current pulse flows through the spin-orbit coupling layer to generate a spin current and the spin current is injected into the ferromagnetic free layer. Under a control of all-electrical controlled, an effective field of a spin-orbit torque drives domain wall to move and displace.

Abstract: A deep neural network training method and apparatus and a computer device are provided. The deep neural network training method includes: obtaining task attributes of nodes in a current network layer in a tree-like network topology (S101); performing cluster analysis on the nodes in the current network layer based on the task attributes of the nodes in the current network layer and extracting a common part of task attributes of multiple nodes in a same category as a task attribute for a parent node of the multiple nodes (S102); training a network parameter of each parent node based on a task attribute of this parent node (S103); and determining that training of a deep neural network corresponding to the tree-like network topology is completed, after completion of training of all nodes in all network layers (S104). The operation efficiency of deep learning can be improved through this solution.

Abstract: The embodiments of the present disclosure provides a method and an apparatus for training a neural network model, A training sample is obtained, and the neural network model is trained using the training sample. When the neural network model is trained, power exponential domain fixed-point encoding is performed on a first activation inputted into each network layer and a network weight of each network layer, and an encoded first activation and an encoded network weight are power exponential domain fixed-point data, which when used in the operation, can cause a matrix multiplication operation involved to be converted into an addition operation in the power exponential domain by means of the power exponential domain encoding. The hardware resources required for the addition operation are significantly less than that required for the multiplication operation, which therefore can greatly reduce the hardware resource overhead required for running the neural network model.

Abstract: Embodiments of the present application provide a data storage method, data acquisition method and device thereof. The method includes allocating an N-dimensional first parameter vector for N pieces of to-be-stored data; performing N-dimensional permutation on the first parameter vector, to obtain N second parameter vectors each having N dimensions; constructing a neural network model that maps the current second parameter vectors to expected data samples of the N pieces of to-be-stored data; adjusting model parameters of the neural network model and/or the first parameter vector until expected data samples of the N pieces of to-be-stored data regress to the N pieces of to-be-stored data, the expected data samples being obtained from the current second parameter vectors based on the trained neural network model; storing the current first parameter vector.

Abstract: Embodiments of methods and apparatuses for object detection and of computer devices are disclosed.

Abstract: Embodiments of the present application provide a method, apparatus, and electronic device for eliminating distortion of a distorted image. Side information components of a distorted image are generated. The distorted image is resulted from image processing on an original image. Side information components represent distortion features of the distorted image with respect to original image. Distorted image color components of the distorted image and the side information components are input into a pre-established convolutional neural network model for convolution filtering to obtain distortion-eliminated image color components. The convolutional neural network model is obtained through training based on a preset training set. The training set comprises original sample images, distorted image color components of multiple distorted images corresponding to each of the original sample images, and side information components of each of the distorted images.

Abstract: A deep neural network operation method and apparatus are provided. The method comprises: obtaining an input feature map of a network layer; displacing respectively, according to a preset displacement parameter, each of channels of the input feature map of the network layer along axes, to obtain a displaced feature map, wherein the preset displacement parameter comprises displacement amounts of the channel in the axes; and performing convolution operation on the displaced feature map with a 1×1 convolution kernel to obtain an output feature map of the network layer. The operation efficiency of the DNN can be improved through the above method.

Abstract: The embodiments of the present application provide an image segmentation method, an image segmentation apparatus, and a fully convolutional network system. The method includes: acquiring a target image to be processed; acquiring image feature data of the target image; inputting the image feature data into a pre-trained target network far image segmentation to obtain an output; wherein the target network is a fully convolutional network comprising a hybrid context network structure, and the hybrid context network structure is configured to extract a plurality of reference features at a predetermined scale and fuse them into a target feature that matches a scale of a target object in a segmented image; and wherein the target network is trained with sample images containing target objects at different scales; and obtaining an image segmentation result for the target image based on the output.

Abstract: An operation method and apparatus for a network layer in a Deep Neural Network are provided.

Abstract: Embodiments of the present application provide a method, apparatus, and electronic device for eliminating distortion of a distorted image. Side information components of a distorted image are generated. The distorted image is resulted from image processing on an original image. Side information components represent distortion features of the distorted image with respect to original image. Distorted image color components of the distorted image and the side information components are input into a pre-established convolutional neural network model for convolution filtering to obtain distortion-eliminated image color components. The convolutional neural network model is obtained through training based on a preset training set. The training set comprises original sample images, distorted image color components of multiple distorted images corresponding to each of the original sample images, and side information components of each of the distorted images.

Abstract: Embodiments of the present application disclose a target detection method and device, and relate to the technical field of video processing.

Abstract: A deep neural network training method and apparatus and a computer device are provided. The deep neural network training method includes: obtaining task attributes of nodes in a current network layer in a tree-like network topology (S101); performing cluster analysis on the nodes in the current network layer based on the task attributes of the nodes in the current network layer and extracting a common part of task attributes of multiple nodes in a same category as a task attribute for a parent node of the multiple nodes (S102); training a network parameter of each parent node based on a task attribute of this parent node (S103); and determining that training of a deep neural network corresponding to the tree-like network topology is completed, after completion of training of all nodes in all network layers (S104). The operation efficiency of deep learning can be improved through this solution.

Abstract: Embodiments of methods and apparatuses for object detection and of computer devices are disclosed.

Abstract: Embodiments of the present application disclose a target detection method and device, and relate to the technical field of video processing.