公开(公告)号：US20230351212A1

公开(公告)日：2023-11-02

申请号：US17837233

申请日：2022-06-10

Applicant: ZHEJIANG LAB

Inventor： Hongsheng WANG ,  Qing LIAO ,  Hujun BAO ,  Guang CHEN

IPC: G06N5/02

CPC classification number: G06N5/022

Abstract: The disclosure provides a semi-supervised method and apparatus for public opinion text analysis. The semi-supervised method includes: first acquiring a public opinion data set, and preprocessing the data set; performing a data augmentation algorithm on preprocessed samples to generate data augmented samples; generating category labels for the unlabeled samples in the data set in an unsupervised extraction and clustering manner; calculating similarities of word vector latent semantic spaces and performing linear interpolation operation to generate, according to an operation result, similarity interpolation samples; constructing a final training sample set; adopting a semi-supervised method, inputting the final training sample set into a pre-trained language model to train the model to obtain a classification model; and predicting the test set by using the classification model to obtain a classification result.

公开(公告)号：US20230315203A1

公开(公告)日：2023-10-05

申请号：US18115678

申请日：2023-02-28

Applicant: ZHEJIANG LAB

Inventor： Yina WEI ,  Lijie WANG ,  Jinbiao LIU ,  Tao TANG ,  Linqing FENG ,  Zhengting CAI

IPC: G06F3/01 ,  A61B5/00 ,  A61B5/378

CPC classification number: G06F3/015 ,  A61B5/7264 ,  A61B5/378

Abstract: The present disclosure discloses a brain-computer interface decoding method and apparatus based on point-position equivalent augmentation. According to the method, a point-position equivalent transformation is performed on sampling points to augment training data and generate arrangement sets. The task-related component analysis is performed on the augmented data to generate spatial filter. Afterwards, a full-frequency directed rearrangement is performed on verification signals or test signals according to the equivalent arrangement sets. After spatial filtering, Pearson correlation coefficients between the rearranged signals and the decoding templates are calculated. These correlation coefficients will be classified and voted by using a naive Bayes method. The verification module will generate the coefficient probability density functions and a threshold, and the test module will finally output the predicted label based on these information.

公开(公告)号：US20230274129A1

公开(公告)日：2023-08-31

申请号：US17706734

申请日：2022-03-29

Applicant: ZHEJIANG LAB

Inventor： Hongsheng WANG ,  Hujun BAO ,  Guang CHEN ,  Lingfang ZENG ,  Hongcai CHENG ,  Yong LI ,  Jian ZHU ,  Huanbo ZHENG

IPC: G06N3/063 ,  G06N3/08

CPC classification number: G06N3/063 ,  G06N3/08

Abstract: The present disclosure discloses a method for execution of a computational graph in a neural network model and an apparatus thereof, including: creating task execution bodies on a native machine according to a physical computational graph compiled and generated by a deep learning framework, and designing a solution for allocating a plurality of idle memory blocks to each task execution body, so that the entire computational graph participates in deep learning training tasks of different batches of data in a pipelining and parallelizing manner.

公开(公告)号：US20230243998A1

公开(公告)日：2023-08-03

申请号：US17927748

申请日：2020-08-28

Applicant: ZHEJIANG UNIVERSITY ,  ZHEJIANG LAB

Inventor： Xingfan CHEN ,  Yishi LIU ,  Xiaowen GAO ,  Nan LI ,  Huizhu HU ,  Cheng LIU

IPC: G01V7/14

CPC classification number: G01V7/14

Abstract: An absolute gravimeter and a measurement method based on vacuum optical tweezers. The micro-nano particle releasing device is equipped with micro-nano particles, and is located above laser optical tweezers, and the laser optical tweezers have two capturing beams which pass through the respective convergent lenses and then converge at an intersection. An area where the intersection is located serves as an optical trap capturing region, and the micro-nano particles are stably captured by the two capturing beams in the optical trap capturing region. The optical interferometer is electrically connected to the signal processing device, the optical interferometer measures a displacement of the micro-nano particles in real time at the beginning of a free fall process from the optical trap capturing region and sends the displacement signal to the signal processing device. The signal processing device obtains a measured value of an absolute gravitational acceleration.

公开(公告)号：US11704773B2

公开(公告)日：2023-07-18

申请号：US17712080

申请日：2022-04-02

Applicant: ZHEJIANG LAB

Inventor： Duo Zhang ,  Wentao Zhu ,  Ling Chen ,  Fan Rao ,  Bao Yang ,  Hui Shen

IPC: G06T3/60 ,  G06T11/00 ,  G06N3/045

CPC classification number: G06T3/60 ,  G06N3/045 ,  G06T11/006 ,  G06T2207/20084 ,  G06T2210/41

Abstract: Disclosed is an automatic reorientation method from an SPECT three-dimensional reconstructed image to a standard view, wherein a rigid registration parameter P between a SPECT three-dimensional reconstructed image A and a standard SPECT image R is extracted by using a rigid registration algorithm to form a mapping database of A and P; features of the image A are extracted by using a three-layer convolution module, and are converted into a 6-dimensional feature vector T after three times of full connection, and T is applied to A through a spatial transformer network to form an orientation result predicted by the network, thus establishing the automatic reorientation model of the SPECT three-dimensional reconstructed image. The SPECT three-dimensional reconstructed image to be orientated is taken as an input. A standard view can be obtained by using the automatic reorientation model of the SPECT three-dimensional reconstructed image for automatic turning.

公开(公告)号：US20230223939A1

公开(公告)日：2023-07-13

申请号：US18183908

申请日：2023-03-14

Applicant: ZHEJIANG LAB

Inventor： Jiani GU ,  Bing CHEN ,  Xiao YU ,  Chengji JIN ,  Genquan HAN

IPC: H03K19/08 ,  H03K19/21

CPC classification number: H03K19/08 ,  H03K19/21 ,  G11C11/223

Abstract: A method, a unit and circuits for implementing Boolean logics based on computing-in-memory transistors. The method is implemented by using the characteristics and the read-write mode of the computing-in-memory transistor; the basic unit consists of a computing-in-memory transistor and a pull resistor; the pull resistor in the basic unit is connected in series with the transistor, and the gate of the transistor is independent; the basic units can implement sixteen Boolean logic operations through different circuit structures and voltage configuration schemes. Compared with the logic circuit structure of the conventional CMOS transistors, the present disclosure can implement more logic operations with fewer transistors, which greatly optimizes circuit density and computing speed caused by data transmission between storage units and process units.

公开(公告)号：US20230191623A1

公开(公告)日：2023-06-22

申请号：US17924696

申请日：2021-11-02

Applicant: ZHEJIANG LAB

Inventor： Lei ZHANG ,  Jianliang XIAO ,  Wenzhen YANG ,  Shuqi MA ,  Ni YAO

IPC: B25J15/00 ,  F03G7/06

CPC classification number: B25J15/00 ,  F03G7/0614 ,  F03G7/0616

Abstract: An optical waveguide-type soft photoactuator based on an optical micro/nanofiber includes an optical micro/nanofiber, a first deformed material membrane, and a second deformed material membrane. One end of the optical micro/nanofiber is provided with a taper region and a waist region, and the taper region and the waist region are encapsulated in the first deformed material membrane. The second deformed material membrane covers a side of the first deformed material membrane, and the first deformed material membrane or the second deformed material membrane is doped with a photothermal conversion material. The refractive index of the first deformed material membrane is less than the refractive index of a core of the optical micro/nanofiber. The coefficient of thermal expansion of the first deformed material membrane and a coefficient of thermal expansion of the second deformed material membrane are different.

公开(公告)号：US20230177312A1

公开(公告)日：2023-06-08

申请号：US17739205

申请日：2022-05-09

Applicant: ZHEJIANG LAB

Inventor： Hongsheng WANG ,  Hujun BAO ,  Wei HUA ,  Weiqiang JIA

IPC: G06N3/04 ,  G06F9/54 ,  G06F9/48 ,  G06F8/36

CPC classification number: G06N3/0454 ,  G06F9/545 ,  G06F9/4881 ,  G06F8/36

Abstract: Disclosed is a method for distributed type training adaptation and apparatus in a deep learning framework and an AI accelerator card. The method includes the following steps: S1: the deep learning framework supports single-card configuration in a newly added AI accelerator card, and sub-steps thereof are as follows: S11: the deep learning framework supports new hardware; S12: the deep learning framework supports a device thread of the new hardware; S13: the deep learning framework supports a memory operation of the new hardware; and S14: the deep learning framework supports an operator kernel function of the new hardware; S2: the deep learning framework supports multi-card configuration in the newly added AI accelerator card; S3: the deep learning framework supports tensor segmentation and multi-card distribution; and S4: the deep learning framework supports multi-card collective communication in the newly added AI accelerator card.

公开(公告)号：US20230169063A1

公开(公告)日：2023-06-01

申请号：US17824349

申请日：2022-05-25

Applicant: ZHEJIANG LAB

Inventor： Peilei WANG ,  Ruyun ZHANG ,  Tao ZOU ,  Peilong HUANG

IPC: G06F16/23 ,  G06F16/27 ,  G06F16/21

CPC classification number: G06F16/2365 ,  G06F16/275 ,  G06F16/21

Abstract: The disclosure discloses a mimetic database-based network operating system design method, including: designing a mimetic data structure; designing a mimetic data object; designing a synchronization mechanism and a decision mechanism, designing a mimetic database safe storage command processing system, and designing a classification storage mechanism for interacting data between service modules and a master database in a network operating system. By means of vertical hierarchy and horizontal classification, the problem of compatibility of the database subjected to mimetic transformation and a network operating system is solved. By means of a memory random distribution storage mechanism and a memory hardware heterogeneous storage mechanism, the cost caused by mimetic transformation can be reduced, and the cost is controllable while the safety is improved.

公开(公告)号：US11645495B2

公开(公告)日：2023-05-09

申请号：US17888513

申请日：2022-08-16

Applicant: Zhejiang Lab

Inventor： Feng Gao ,  Wenyuan Bai

IPC: G06N3/04 ,  G06N3/08

CPC classification number: G06N3/04 ,  G06N3/08

Abstract: The present invention discloses an edge calculation-oriented reparametric neural network architecture search method, including the following steps: S1: designing linear operators and multi-branch block structures; S2: constructing a hypernetwork by stacking the multi-branch block structures; S3: training the hypernetwork through a gradient-based first-stage search algorithm; S4: deleting redundant branches in the hypernetwork to construct an optimal subnetwork; S5: converting the multi-branch optimal subnetwork into a single-branch network; and S6: completing task reasoning by using the single-branch network. The method is used to search the neural network structure capable of performing reparameterization, and ensures the reasoning real-time performance and the high efficiency of model operation while ensuring the reasoning precision.