Abstract: The present disclosure provides a multi-modal multi-view classification method for echocardiograms based on a deep learning algorithm, including: collecting videos and images of multi-modal multi-view adult echocardiograms, and preprocessing the videos and images; annotating the preprocessed videos and images of adult echocardiograms to generate an adult echocardiogram dataset; dividing the adult echocardiogram dataset into a training set, a validation set, and a test set; constructing an adult echocardiogram view classification model based on a ResNet network, training the model using the training set, and selecting an optimal classification model using the validation set; evaluating performance of the optimal adult echocardiogram view classification model based on the test set; and inputting a to-be-tested image or video into the adult echocardiogram view classification model to obtain a classification result.

Abstract: Provided is a fault signal locating and identifying method of industrial equipment based on a microphone array. The method includes the steps of: acquiring sound signals and dividing the acquired signals into a training set, a verifying set and a test set; performing feature extraction on the sound signals in the training set, and extracting a phase spectrogram and an amplitude spectrogram of a spectrogram; sending an output of a feature extraction module, as an input, to a CNN, and in each layer of the CNN, learning a translation invariance in the spectrogram by using a 2D CNN; in between the layers of the CNN, normalizing the output by using a batch normalization, and reducing a dimension by using a maximum pooling layer along a frequency axis; sending an output from the layers of the CNN to layers of RNN; using a linear activation function; and inputting an output of a full connection layer to two parallel full connection layer branches for fault identification and fault location, respectively.

Abstract: Provided is a fault signal locating and identifying method of industrial equipment based on a microphone array. The method includes the steps of: acquiring sound signals and dividing the acquired signals into a training set, a verifying set and a test set; performing feature extraction on the sound signals in the training set, and extracting a phase spectrogram and an amplitude spectrogram of a spectrogram; sending an output of a feature extraction module, as an input, to a CNN, and in each layer of the CNN, learning a translation invariance in the spectrogram by using a 2D CNN; in between the layers of the CNN, normalizing the output by using a batch normalization, and reducing a dimension by using a maximum pooling layer along a frequency axis; sending an output from the layers of the CNN to layers of RNN; using a linear activation function; and inputting an output of a full connection layer to two parallel full connection layer branches for fault identification and fault location, respectively.

Abstract: The present invention discloses an active extension type car front bumper. Three fixed sleeves are rigidly connected to a cross beam of a U-shaped fixed frame, an impact rod is fitted in each fixed sleeve, a front end of each impact rod passes through the cross beam and is fixedly connected to the bumper, and a rear end of each impact rod is fixedly connected to a movable cross beam arranged horizontally left and right. Left and right longitudinal beams of the U-shaped fixed frame are each provided with a movable longitudinal beam that is movable forward and backward. Left and right ends of the movable cross beam respectively extend into the left and right longitudinal beams. A rear end of each movable longitudinal beam is fixedly connected to a hydraulic mechanism. An energy absorption mechanism is provided between every two adjacent fixed sleeves.

Abstract: The present invention discloses an active extension type car front bumper. Three fixed sleeves are rigidly connected to a cross beam of a U-shaped fixed frame, an impact rod is fitted each fixed sleeve, a front end of each impact rod passes through the cross beam and is fixedly connected to the bumper, and a rear end of each impact rod is fixedly connected to a movable cross beam arranged horizontally left and right. Left and right longitudinal beams of the U-shaped fixed frame are each provided with a movable longitudinal beam that is movable forward and backward. Left and right ends of the movable cross beam respectively extend into the left and right longitudinal beams. A rear end of each movable longitudinal beam is fixedly connected to a hydraulic mechanism. An energy absorption mechanism is provided between every two adjacent fixed sleeves.

Abstract: A model-adaptive multi-source large-scale mask projection 3D printing system configured to conduct the following steps: projecting pure-color images of first and second colors having identical attributes, capturing an image of an overlapping portion and calculating height and width information of the overlapping portion; splitting a pre-processed slice and respectively recording width and height information of two slices resulting from the splitting and generating two gray scale images having identical attributes thereto; counting power values of identical positions of slices in different gray scale values, performing a further calculation to obtain a projection mapping function, using the projection mapping function as a basis for performing optimization on gray scale interpolation of the generated images; and fusing the processed gray scale images and the originally split two slices to obtain a mask projection 3D printing slice having a uniform shaping brightness, and forming a final product.

Abstract: A multi-degree-of-freedom (Multi-DOF) 3D printing device includes a printer mechanism, a rotatable platform, a visual inspection system, and a control system. The printing mechanism includes a print head. The printing mechanism and the rotatable platform are combined to have three degrees of freedom of translation, and the rotatable platform has two degrees of freedom of rotation. The visual inspection system includes a camera mounted to nearby the print head and the relative position of camera and print head remains constant. The control system is configured to implement the printing process. The model is decomposed into several components, each of which could be printed in a single direction. After one component is printed out, the platform is then rotated so that the cutting plane for the component will be printed is horizontal.

Abstract: A computer-based system trains a neural network by solving a double layer optimization problem. The system includes an input interface to receive an input to the neural network and labels of the input to the neural network; a processor to solve a double layer optimization to produce parameters of the neural network, and an output interface to output the parameters of the neural network. The double layer optimization includes an optimization of a first layer subject to an optimization of a second layer. The optimization of the first layer minimizes a difference between an output of the neural network processing the input and the labels of the input to the neural network, the optimization of the second layer minimizes a distance between a non-negative output vector of each layer and a corresponding input vector to each layer. The input vector of a current layer is a linear transformation of the non-negative output vector of the previous layer.

Abstract: A model-adaptive multi-source large-scale mask projection 3D printing system configured to conduct the following steps: projecting pure-color images of first and second colors having identical attributes, capturing an image of an overlapping portion and calculating height and width information of the overlapping portion; splitting a pre-processed slice and respectively recording width and height information of two slices resulting from the splitting and generating two gray scale images having identical attributes thereto; counting power values of identical positions of slices in different gray scale values, performing a further calculation to obtain a projection mapping function, using the projection mapping function as a basis for performing optimization on gray scale interpolation of the generated images; and fusing the processed gray scale images and the originally split two slices to obtain a mask projection 3D printing slice having a uniform shaping brightness, and forming a final product.

Abstract: A light homogenization method for multi-source large-scale surface exposure 3D printing, comprising the following steps: projecting pure-color images of a first color and a second color having identical attributes capturing an image of an overlapping portion and calculating height and width information of the overlapping portion; splitting a pre-processed slice and respectively recording width and height information of two slices resulting from the splitting and generating two grayscale images having identical attributes thereto; counting power values of identical positions of slices in different grayscale values, performing a further calculation to obtain a projection mapping function, using the projection mapping function as a basis for performing optimization on grayscale interpolation of the generated images; and fusing the processed grayscale images and the originally split two slices to obtain a surface exposure 3D printing slice having a uniform brightness in final shaping.

Abstract: A multi-degree-of-freedom (Multi-DOF) 3D printing device includes a printer mechanism, a rotatable platform, a visual inspection system, and a control system. The printing mechanism includes a print head. The printing mechanism and the rotatable platform are combined to have three degrees of freedom of translation, and the rotatable platform has two degrees of freedom of rotation. The visual inspection system includes a camera mounted to nearby the print head and the relative position of camera and print head remains constant. The control system is configured to implement the printing process. The model is decomposed into several components, each of which could be printed in a single direction. After one component is printed out, the platform is then rotated so that the cutting plane for the component will be printed is horizontal.

Abstract: A light homogenization method for multi-source large-scale surface exposure 3D printing, comprising the following steps: projecting pure-color images of a first color and a second color having identical attributes, capturing an image of an overlapping portion and calculating height and width information of the overlapping portion; splitting a pre-processed slice and respectively recording width and height information of two slices resulting from the splitting and generating two grayscale images having identical attributes thereto; counting power values of identical positions of slices in different grayscale values, performing a further calculation to obtain a projection mapping function, using the projection mapping function as a basis for performing optimization on grayscale interpolation of the generated images; and fusing the processed grayscale images and the originally split two slices to obtain a surface exposure 3D printing slice having a uniform brightness in final shaping.

Abstract: A computer-based system trains a neural network by solving a double layer optimization problem. The system includes an input interface to receive an input to the neural network and labels of the input to the neural network; a processor to solve a double layer optimization to produce parameters of the neural network, and an output interface to output the parameters of the neural network. The double layer optimization includes an optimization of a first layer subject to an optimization of a second layer. The optimization of the first layer minimizes a difference between an output of the neural network processing the input and the labels of the input to the neural network, the optimization of the second layer minimizes a distance between a non-negative output vector of each layer and a corresponding input vector to each layer. The input vector of a current layer is a linear transformation of the non-negative output vector of the previous layer.

Abstract: One embodiment of the present solution is a relationship-based video server system. The server system is programmed to provide a platform for large scale video search based on relationships between objects in video segments, and the platform executes search based on probabilistically discriminative objects and relationships.

Abstract: One embodiment of the present solution is a relationship-based video server system. The server system is programmed to provide a platform for large scale video search based on relationships between objects in video segments, and the platform executes search based on probabilistically discriminative objects and relationships.

Abstract: The present invention provides compounds having the general structure I, or a pharmaceutically acceptable salt thereof: wherein Z is selected from a group consisting of O, C, N, and S; each of Ar1, Ar2, and Ar3 is a moiety selected from a group consisting of an aryl and a substituted aryl; Y is selected from a group consisting of S, O, CH2 and SO2; K is selected from a group consisting CH2, CF2, O, NH, and SO2; each of n1 and n2 is an integer selected from a group consisting of 0, 1, 2, 3, and 4; and each of x, y, m1 and m2 is an integer selected from a group consisting of 0 and 1. Pharmaceutical compositions based on compounds I are also provided for treatment of a neurological disorders, diseases, or pathologies, and for treatment of infectious diseases.