Method for machine reading comprehension

    公开(公告)号:US11620449B2

    公开(公告)日:2023-04-04

    申请号:US17024726

    申请日:2020-09-18

    摘要: A method for machine reading comprehension includes: S1, obtaining a character-level indication vector of a question and a character-level indication vector of an article; S2, obtaining an encoded question vector and an encoded article vector; S3, obtaining an output P1 of a bidirectional attention model and an output P2 of a shared attention model; S4, obtaining an aggregated vector P3; S5, obtaining a text encoding vector P4; S6, obtaining global interaction information between words within the article; S7, obtaining a text vector P5 after using the self-attention model; S8, obtaining aggregated data P6 according to the text encoding vector P4 and the text vector P5; S9, obtaining a context vector of the article according to the aggregated data P6 and an unencoded article vector P; and S10, predicting an answer position according to the context vector of the article and the encoded question vector to complete the machine reading comprehension.

    IN-SITU VACUUM REACTION SYSTEM FOR DYNAMICALLY DETECTING DEFECTS

    公开(公告)号:US20230051210A1

    公开(公告)日:2023-02-16

    申请号:US17881794

    申请日:2022-08-05

    IPC分类号: G01N24/10

    摘要: Disclosed is an in-situ vacuum reaction system for dynamically detecting defects, which includes an electron paramagnetic resonance spectrometer, an in-situ vacuum reaction chamber, a gas supply unit, a vacuum unit, an illumination unit, a temperature control unit and a mixing bottle, the in-situ vacuum reaction chamber is arranged inside a detection cavity of the electron paramagnetic resonance spectrometer, the gas supply unit is connected to the mixing bottle through a pipeline, and the mixing bottle is connected to a gas inlet of the in-situ vacuum reaction chamber through a pipeline, the vacuum unit is connected to the gas inlet of the in-situ vacuum reaction chamber through a pipeline, the illumination unit is arranged corresponding to a detachable window of the electron paramagnetic resonance spectrometer, and the temperature control unit is connected to the electron paramagnetic resonance spectrometer through a pipeline.

    WIRELESS SINGLE-PHASE AC-TO-AC CONVERSION CIRCUIT BASED ON 2.4G MICROWAVE

    公开(公告)号:US20220385323A1

    公开(公告)日:2022-12-01

    申请号:US17645318

    申请日:2021-12-21

    摘要: A wireless single-phase AC-to-AC conversion circuit based on a 2.4G microwave includes a receiving antenna unit, a RF switch unit, a positive voltage rectification unit, a negative voltage rectification unit and an AC synthesis unit. An output port of the receiving antenna unit is connected to the common input port of the RF switch unit. A first microwave output end of the RF switch unit and a second microwave output end of the RF switch unit are correspondingly connected to a microwave input end of the positive voltage rectification unit and a microwave input end of the negative voltage rectification unit, respectively. ADC output end of the positive voltage rectification unit and a DC output end of the negative voltage rectification unit are correspondingly connected to a positive voltage input port of the AC synthesis unit and a negative voltage input port of the AC synthesis unit, respectively.

    An Auxiliary Diagnostic Model and an Image Processing Method for Detecting Acute Ischemic Stroke

    公开(公告)号:US20220148301A1

    公开(公告)日:2022-05-12

    申请号:US17606751

    申请日:2020-09-29

    IPC分类号: G06V10/82 G06T3/40 G06N3/04

    摘要: This invention discloses an auxiliary diagnostic model and an image processing method for detecting acute ischemic stroke. This refers to the technical field of medical image processing. The technical essentials are described as follow: the presented deep-learning model is based on generative adversarial networks (GANs), comprising a generator (G) and a discriminator (D). G is the first three-dimensional convolutional neural network, used to synthesize realistic images from raw data, while D is the second three-dimensional convolutional neural network, used to classify images as real or fake (synthetic). The presented GAN model can learn the mapping relationship from non-enhanced computed tomography (NECT) images to T2-weighted fluid-attenuation inversion recovery (FLAIR) magnetic resonance imaging (MRI), then converting the raw CT to synthetic FLAIR with high sensitivity. This improves the efficiency of emergency scanning in acute ischemic stroke, reaching sensitivity that is poor in CT interpretation and immediacy that is limited in MRI examination.

    FLEXIBLE MICROWAVE POWER TRANSISTOR AND PREPARATION METHOD THEREOF

    公开(公告)号:US20220102542A1

    公开(公告)日:2022-03-31

    申请号:US17264521

    申请日:2019-12-20

    摘要: The present disclosure provides a flexible microwave power transistor and a preparation method thereof. In view of great lattice mismatch and poor performance of a device prepared with a Si substrate in an existing preparation method, the preparation method of the present disclosure grows a gallium nitride high electron mobility transistor (GaN HEMT) layer on a rigid silicon carbide (SiC) substrate to avoid lattice mismatch between a silicon (Si) substrate and gallium nitride (GaN), improving performance of the flexible microwave power transistor. Moreover, in view of problems such as low output power, power added efficiency and power gain with the existing device preparation method, the present disclosure retains part of the rigid SiC substrate and grows a flexible substrates at room temperature to prepare a high-quality device. The present disclosure has greatly improved power output capability, efficiency and gain, and basically unchanged performance of device under 0.75% of stress.