Patents by Inventor Hongsheng Chen

Hongsheng Chen has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).

  • Publication number: 20240104341
    Abstract: A memory optimization method includes: compiling a neural network into a computational graph for neural network computation on a computer; transforming the computational graph into a topological graph; constructing a life cycle relationship graph of tensor variables in the computational graph; and analyzing a life cycle relationship among tensor variables in a node of the computational graph; iteratively merging those tensor variables connected by lines of the second type and caching into a memory any tensor variable that goes beyond a number of idle registers and is not allocated to a register, until all tensor variables that go beyond the number of the idle registers and are not allocated to registers are cached into the memory; caching any node of the life cycle relationship graph with a degree smaller than a number of registers into a stack.
    Type: Application
    Filed: November 22, 2022
    Publication date: March 28, 2024
    Inventors: Hongsheng WANG, Guang CHEN, Lingfang ZENG
  • Publication number: 20240104395
    Abstract: Disclosed are a memory optimization method and device oriented to neural network computing. The memory optimization method oriented to neural network computing includes the following steps: step S1: reconstructing a computation graph into a topological structure computation graph; step S2: constructing a life cycle interval about tensor variables; step S3: constructing a scanning line about the life cycle interval; step S4: allocating the tensor variables to idle registers; step S5: allocating to tensor variables exceeding the required number of registers; step S6: allocating registers allocated in the expired life cycle interval to tensor variables exceeding the required number of registers; and step S7: adding tensor variables transferred to a memory back to the life cycle interval in an activated state, and allocating idle registers for the tensor variables. According to the present disclosure, the memory of a data flow of a computation graph for neural network computing is optimized.
    Type: Application
    Filed: December 1, 2022
    Publication date: March 28, 2024
    Inventors: Hongsheng WANG, Guang CHEN
  • Publication number: 20240104016
    Abstract: The disclosure discloses an intermediate representation method for compiling computation graphs, including: step 1: compiling a neural network into a computation graph for neural network computation; step 2: constructing a node for each tensor variable in the computation graph; step 3: associating the node representing the tensor variable in the computation graph to a set of pointers to the tensor variable; step 4: analyzing constraint relationships between the tensor variables in the computation graph; step 5: iteratively constructing a topological graph of the intermediate representation based on the constraint relationships between the tensor variables in the computation graph; and step 6: analyzing the tensor variables with different aliases pointing to a same memory location based on the intermediate representation, and allocating a register for the tensor variables with different aliases.
    Type: Application
    Filed: November 30, 2022
    Publication date: March 28, 2024
    Inventors: Hongsheng WANG, Aimin PAN, Guang CHEN
  • Patent number: 11941514
    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.
    Type: Grant
    Filed: March 29, 2022
    Date of Patent: March 26, 2024
    Assignee: ZHEJIANG LAB
    Inventors: Hongsheng Wang, Hujun Bao, Guang Chen, Lingfang Zeng, Hongcai Cheng, Yong Li, Jian Zhu, Huanbo Zheng
  • Patent number: 11941507
    Abstract: Disclosed are a data flow method and apparatus for neural network computation. The data flow method for neural network computation includes initializing the lifecycle of a variable in a computational graph; and defining a propagation rule for a variable in use to flow through a node. A definition of the variable is produced at a precursor node of the node, such that an input set of valid variables flowing through the node contains the variable. The method may be used on neural network computation in a deep learning training system.
    Type: Grant
    Filed: September 27, 2022
    Date of Patent: March 26, 2024
    Assignee: ZHEJIANG LAB
    Inventors: Hongsheng Wang, Guang Chen
  • Patent number: 11934887
    Abstract: The present disclosure discloses a distributed model compilation system. A master node of the system determines the logic calculation graph of the model based on model information, divides the logic calculation graph into multiple logic calculation sub-graphs, generates a distributing message for each logic calculation sub-graph, and then transmits the distributing message to a slave node. Each of the slave nodes allocates a local computing resource to compile the logic calculation sub-graph based on the received distributing message, and transmits compilation completion information to the master node. The master node determines the completion of model compilation based on the compilation completion information returned by each slave node, and executes the target work based on the compiled model.
    Type: Grant
    Filed: September 13, 2023
    Date of Patent: March 19, 2024
    Assignee: ZHEJIANG LAB
    Inventors: Hongsheng Wang, Fei Wu, Guang Chen, Feng Lin
  • Patent number: 11915135
    Abstract: The disclosure discloses a graph optimization method and apparatus for neural network computation. The graph optimization method includes the following steps: S1: converting a computation graph; S2: allocating a register; S3: defining a route selector for a redefined variable; S4: solving the route selector for the redefined variable; S5: defining a criterion of inserting the route selector for the redefined variable into a node; S6: analyzing a dominating edge set of the node for the redefined variable; S7: inserting the route selector for the redefined variable; and S8: renaming the redefined variable. The disclosure solves the problem of the corresponding route selection on a correct definition of the redefined variable when a node including the redefined variable in a computation graph in the compiling period flows through multiple paths of computation flow, reduces the memory cost and promotes the development of implementation application of a deep neural network model.
    Type: Grant
    Filed: September 21, 2022
    Date of Patent: February 27, 2024
    Assignee: ZHEJIANG LAB
    Inventors: Hongsheng Wang, Guang Chen
  • Publication number: 20220110491
    Abstract: A tooth-brushing cup and heating device consists of a tooth-brushing cup, a hoop, a base and the like an electromagnetic coil, a control circuit and the like are arranged in the base, a display window is arranged at the upper part of the base. Four buttons are arranged above the display window for setting the temperature and the heating time respectively, and a power switch is arranged on the right side of the display window. The hoop is installed on the base, and a conical surface inside the hoop is attached to the outer conical surface of the tooth-brushing cup. The tooth-brushing cup is made of stainless steel by stamping and is in a circular table shape, and a side plane is stamped on the side wall of the tooth-brushing cup to be tightly attached to a base panel.
    Type: Application
    Filed: June 9, 2020
    Publication date: April 14, 2022
    Inventor: Hongsheng Chen
  • Publication number: 20210342658
    Abstract: A polysemant meaning learning method is provided. The method includes extracting a plurality of first target terms and at least one adjacent term combinations of each first target term; obtaining a capsule network model by training by taking the encoding of each first target term as an input vector and the encoding of each adjacent term combination corresponding to each first target term as an output vector; when a to-be-recognized second target term is recognized, inputting the second target term into the capsule network model, and determining a plurality of obtained intermediate vectors as feature vectors of the second target term; and clustering the feature vectors with a cosine similarity greater than a similarity threshold to generate representative terms of one or more categories and determining one or more meanings of the one or more categories.
    Type: Application
    Filed: July 30, 2019
    Publication date: November 4, 2021
    Inventor: Hongsheng CHEN
  • Publication number: 20200240680
    Abstract: A heat pump unit and methods for operating the heat pump unit are provided. The heat pump unit includes a compressor (101), a throttling device (107), a first heat exchanger (104), a second heat exchanger (102), a third heat exchanger (103) and a mid-pressure tank (110). The heat pump unit operates in multiple run modes and switches between the run modes without shutdown. The first heat exchanger or the second heat exchanger is capable of acting as a condenser in the multiple run modes. When switching from a pre-switching run mode to a post-switching run mode, a control device determines whether to perform a pressure release operation to the first heat exchanger or the second heat exchanger.
    Type: Application
    Filed: September 28, 2018
    Publication date: July 30, 2020
    Inventors: Lue Lv, Hongsheng Chen, Qian Hu
  • Publication number: 20200132491
    Abstract: The present application provides a method and an apparatus for generating POI copy. According to an example of the method, after POI information for generating copy is obtained, topic trend prediction may be performed on the POI information to determine a topic word of the POI information, and POI copy corresponding to the POI information is generated based on the topic word.
    Type: Application
    Filed: December 20, 2017
    Publication date: April 30, 2020
    Inventors: Hua ZHANG, Changlin DING, Hongsheng CHEN
  • Patent number: 9971161
    Abstract: An electromagnetic wave cloaking device is described. The device may include a plurality of first-type dielectric units, a plurality of second-type dielectric units, and a plurality of four kinds of spacers. The dielectric units and the spacers are arranged such that an electromagnetic wave incident into the cloaking device can be refracted within the device and around a cloaking area surrounded by the device.
    Type: Grant
    Filed: March 10, 2013
    Date of Patent: May 15, 2018
    Assignee: ZHEJIANG UNIVERSITY
    Inventors: Hongsheng Chen, Bin Zheng, Baile Zhang
  • Publication number: 20170240294
    Abstract: A three-dimensional infrared laser aircraft landing-guiding system with high directivity is described. The system utilizes a highly penetrable and highly directional infrared laser source to generate an optical image via an optical image generating apparatus. The generated optical image involves the information for aircraft landing, and the information covers a large range of flying areas to make the optical image detectable for an aircraft. With the information provided by the optical image, the aircraft can be guided to the best landing path. The infrared laser source is highly directional and can emit a laser light to specific directions, reducing waste of energy, increasing navigation distances, and improving anti-jamming capability.
    Type: Application
    Filed: April 13, 2015
    Publication date: August 24, 2017
    Inventors: Hongsheng Chen, Zhaoyun Duan, Min Chen, Tsong-Shin Lim
  • Publication number: 20150248013
    Abstract: An electromagnetic wave cloaking device is described. The device may include a plurality of first-type dielectric units, a plurality of second-type dielectric units, and a plurality of four kinds of spacers. The dielectric units and the spacers are arranged such that an electromagnetic wave incident unto the cloaking device can be refracted within the device and around a cloaking area surrounded by the device.
    Type: Application
    Filed: March 10, 2013
    Publication date: September 3, 2015
    Inventors: Hongsheng Chen, Bin Zheng, Baile Zhang
  • Patent number: 6556199
    Abstract: The present invention relates to an apparatus and method for voxelizing geometric and volumetric models of objects in 3D space, i.e. converting these 3D models into a voxel-based volume representation. Based on an unconventional way of using the 3D graphics hardware in modern graphics workstations, the present invention generates slices of the model as frame buffer images, and then stores the slices in a 3D texture memory for interactive volume rendering. Since most of the computation is carried out directly by hardware, the voxelization algorithms of the present invention are much faster than existing algorithms, and are able to achieve interactive volume rendering and manipulation of complex 3D scenes consisting of intermixed geometric models and sampled volume data sets (e.g. CT and MRI scans).
    Type: Grant
    Filed: August 11, 1999
    Date of Patent: April 29, 2003
    Assignee: Advanced Research and Technology Institute
    Inventors: Shiaofen Fang, Hongsheng Chen