Abstract: A system is programmed to build a machine learning model that comprises a series of masking blocks. Each masking block receives a certain feature vector of an audio segment. Each masking block comprises a first component that generates a first mask for extracting clean speech and a second component that generates a second mask for extracting residual speech masked by the first mask. Each masking block also generates a specific feature vector based on the first mask and the second mask, which becomes the certain feature vector for the next masking block. The second component, which may comprise a gated recurrent unit layer, is computationally less complex than the first component, which may comprise multiple convolutional layers. Furthermore, the system is programmed to receive an input feature vector of an input audio segment and execute the machine learning model to obtain an output feature vector of an output audio segment.

Abstract: Provided are a vegetable oil production device and a vegetable oil production method. The device includes a raw material tank, a heat exchanger and a membrane structure. The membrane structure includes multiple membrane assemblies which are in communication in sequence, and each membrane assembly includes multiple membrane tubes which are arranged in parallel. The raw material tank is provided with a feed port, a discharge port of the raw material tank communicates with a feed port of the membrane structure and a sewage pipe, a clear liquid outlet of the membrane structure communicates with a collection structure, a feed port of the heat exchanger communicates with a discharge port of the membrane structure, a discharge port of the membrane structure further communicates with the sewage pipe, and a discharge port of the heat exchanger communicates with a return port of the raw material tank.

Abstract: A method for enhancing audio signals is provided. In some implementations, the method involves (a) obtaining a training set comprising a plurality of training samples, each training sample comprising a distorted audio signal and a clean audio signal. In some implementations, the method involves (b), for a training sample of the plurality of training samples: obtaining a frequency-domain representation of the distorted audio signal; providing the frequency-domain representation to a convolutional neural network (CNN) comprising a plurality of convolutional layers and to a recurrent element, wherein an output of the recurrent element is provided to a subset of the plurality of convolutional layers; generating a predicted enhancement mask, wherein the CNN generates the predicted enhancement mask; generating a predicted enhanced audio signal based on the predicted enhancement mask; and updating weights associated with the CNN and the recurrent element based on the predicted enhanced audio signal.

Abstract: The disclosure discloses a precipitation soil erosion and microbial respiration monitoring device, which belongs to the technical field of monitoring devices, comprising a precipitation regulation unit, an erosion collection unit, a precipitation monitoring unit, and a microbial respiration monitoring unit. The precipitation monitoring unit is arranged on the side of the precipitation regulation unit, the erosion collection unit is arranged below the side of the precipitation regulation unit, the precipitation regulation unit is configured to adjust the precipitation to simulate the erosion of different precipitation on the soil, and the microbial respiration monitoring unit is configured to monitor the microbial respiration in the soil. The disclosure uses the above device to realize the simultaneous monitoring of soil erosion and microbial respiration, and provides basic data for an in-depth understanding of soil environmental dynamics.

Abstract: At least one processor identifies dependency relationships among libraries in a repository of libraries. Using the dependency relationships among libraries, at least one machine learning model can be created that predicts with a confidence value a dependency between a given library and a target library. An L layer tree-like graph can be created, using the dependency relationships among libraries and an application package. L can be configurable. Versions of the libraries to use can be determined by running the at least one machine learning model for each pair of nodes having a dependency relationship in the L layer tree-like graph, the at least one machine learning model identifying the dependency relationship with a confidence value, where pairs of nodes having largest confidence values are selected as the versions of the libraries to use in the application package.

Abstract: An approach is provided in which the approach loads a machine learning model and a set of test case statistical data into a user system. The set of test case statistical data is based on a set of test cases corresponding to the machine learning model and includes a plurality of input parameter sets and a corresponding set of output quality measurements. The approach compares user data on the user system against the set of test case statistical data and identifies one of the plurality of input parameter sets to optimize the machine learning model based on the set of output quality measurements. The approach generates an optimized machine learning model using the machine learning model and the identified input parameter set at the user system.

Abstract: Method and system for generating a set of synthesized AIRs from a real acoustic impulse response, AIR(t), and using the set of synthesized AIRs to train a machine learning model, such that the machine learning model, after training, is configured to generate a de-reverberated audio signal given an input audio signal. The synthesized AIRs are generated by forming an early portion, AIRe(t) and a late portion, AIRl(t) of the real AIR by selecting a random separation time point, s, and a random crossfade duration, d. With the proposed approach, a “soft” separation of the real AIR into an early AIR and a late AIR. Specifically, the early AIR will decay to zero during a transition period d, while the late AIR will gradually increase from zero during the transition period. The sum of the early AIR and late AIR will still be equal to the real AIR.

Abstract: A method for task processing, an electronic apparatus and a storage medium are provided. The method includes: determining a first computing density level for processing one task or a plurality of tasks; gradually increasing a computing density of a computing unit for processing the one task or the plurality of tasks to the first computing density level within a first time period before processing the one task or the plurality of tasks; processing the one task or the plurality of tasks by the computing unit; and gradually reducing the computing density of the computing unit to a second computing density level within a second time period after completion of processing the one task or the plurality of tasks.

Abstract: Described herein is a method of performing content-aware audio processing for an audio signal comprising a plurality of audio components of different types. The method includes source separating the audio signal into at least a voice-related audio component and a residual audio component. The method further includes determining a dynamic audio gain based on the voice-related audio component and the residual audio component. The method also includes performing audio level adjustment for the audio signal based on the determined audio gain. Further described are corresponding apparatus, programs, and computer-readable storage media.

Abstract: Methods and systems of training a machine learning model include identifying an object or person related to an action in a first video. The object or person is copied from the first video to a second video to generate a third video. A machine learning model is trained using the first video and the third video.

Abstract: A six-roller spreading mechanism and a laminator having the same. The laminator comprises a new six-roller spreading mechanism, which comprises a metering roller, a transfer steel roller, a transfer rubber roller, a spreading steel roller, a spreading rubber roller and a spreading pressure roller, wherein the spreading rubber roller is movably connected to a base by means of a linear motion mechanism so as to enable the spreading rubber roller to linearly move on the base. The new six-roller spreading mechanism is used by the laminator. The active linear motion mechanism is connected to the spreading rubber roller to enable the spreading rubber roller to move linearly, such that control over an acting force between the spreading rubber roller and the spreading pressure roller and the spreading steel roller is achieved, thereby solving the problem of slipping caused by using a low-viscosity glue, also solving the additional problem of the feeding amount of the glue, and thus improving the spreading quality.

Abstract: This application provides a display substrate, a method for manufacturing the same, and a display device, which belongs to the technical field of displays. The method for manufacturing the display substrate comprises: forming a metal layer on a base substrate; forming a photoresist layer on the metal layer; exposing the photoresist layer by using a mask plate; developing the exposed photoresist layer to form a photoresist pattern; and omitting a step of baking the photoresist pattern, and etching the metal layer by using the photoresist pattern as a mask to form a signal line. The technical solution of the present disclosure can realize a narrow frame of a display product.

Abstract: The present application relates to the technical field of electronic materials, in particular to a metal complex and an organic electroluminescent device. The metal complex provided by the present application has the following structure: and thus the organic electroluminescent device containing the metal complex may have high external quantum efficiency and a long device life.

Abstract: This application provides a screen brightness adjustment method, an electronic device, and a storage medium. The method may be applied to an electronic device, and the electronic device includes a screen and an ambient light sensor. The method includes: receiving a first preset operation of turning on the screen by a user; in response to the first preset operation, collecting first light intensity data by using the ambient light sensor, where the first light intensity data is used to update backlight brightness of the screen; turning on the screen by using first backlight brightness corresponding to the first light intensity data; and periodically collecting, after the screen is turned on, second light intensity data by using the ambient light sensor, and adjusting the backlight brightness of the screen by using second backlight brightness corresponding to the second light intensity data.

Abstract: Embodiments are disclosed for a method. The method includes determining a correlation list of missing value predictors. The method also includes generating a cluster model having multiple clusters. The cluster model is based on a target value and predictor values. The method further includes determining an imputed value for a missing value of a row of original training data based on a linear regression model for multiple non-missing value predictor values for the clusters.

Abstract: The present application belongs to the technical field of medical model equipment, and specifically relates to a flexible spine model and a demonstration and test system implemented based on the same. The spine model includes hard structures and a soft elastic connecting material, and the hard structures simulate a layout of spine curvatures and are connected into a whole through the soft connecting material. 1) The present application provides a spine model closer to mechanical characteristics of a human spine and can provide a device for practicing and teaching spine correction treatment for learners. 2) The present application provides a scoliosis correction demonstration model, which can provide a whole visual process of scoliosis correction demonstration for the learners to help improve the learning effect. 3) The present application provides a spine internal fixation system test model which can be used for evaluating properties of different spine internal fixation systems.

Abstract: Disclosed herein is a system for implementing a management controller on a node, or network server, that is dedicated to monitoring the individual health of a plurality of accelerator modules configured on the node. Based on the monitored health, the management controller is configured to implement autonomous power cycle control of individual accelerator modules. The autonomous power cycle control is implemented without violating the requirements of standards established for accelerator modules (e.g., OPEN COMPUTE PROJECT requirements, PERIPHERAL COMPONENT INTERCONNECT EXPRESS (PCIe) interface requirements).

Abstract: The disclosed systems and methods assist transmitted content in reaching a receiver, even if a network is subjected to sophisticated tampering. A device for a communication network including a plurality of nodes and a plurality of paths connecting a source node, which is a transmission source of information, and a terminal node, which is a destination of the information. A segmentation instruction unit being configured to instruct the source node to disperse the information into a plurality of pieces of random number data, to encode the random number data by an error correction code to generate codewords, and to order the codewords from the beginning and divide the codewords into a plurality of segments, the first transmission instruction unit being configured to instruct the source node to transmit OTP-encrypted data of the plurality of segments through the plurality of paths.

Abstract: Embodiments of the present disclosure relate to an image processing method and apparatus, a device, and a medium. The method includes: acquiring an original image to be processed and a perspective correction direction of the original image, wherein the perspective correction direction includes a preset direction and/or a custom direction; acquiring image transformation information corresponding to the original image according to the perspective correction direction, wherein the image transformation information includes one or more of image scaling information, image rotation information, or target point position information; and obtaining a corrected image by performing correction processing on the original image based on the image transformation information.