-
公开(公告)号：US12242477B2
公开(公告)日：2025-03-04
申请号：US17467516
申请日：2021-09-07
Applicant: International Business Machines Corporation
Inventor： Teng Sun , Tong Liu , Si Tong Zhao , XueLiang Zhao , Frank Feng , Yu Zui Wy You , Zhong Fang Yuan
IPC: G06F16/2453 , G06F16/21 , G06F16/242 , G06F16/901 , G06N20/00
Abstract: In order to perform a semantic search based on a graph database, sets of nodes are selected from a plurality of nodes in a graph database. A set of nodes semantically matches a keyword in a natural language query. At least one target node is identified in the sets of nodes. A path is selected from candidate paths based on similarities between the candidate paths and a plurality of paths in the graph database. A graph query for retrieving information from the graph database is generated based on the selected path and the query target.
-
公开(公告)号：US20200302350A1
公开(公告)日：2020-09-24
申请号：US16356643
申请日：2019-03-18
Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATION
Inventor： Teng Sun , Ying Wei , Ting Yao Liu , Yuan Li , Qiang He
Abstract: A method, computer system, and computer program product for NLP-based domain modeling are provided. The embodiment may include receiving, by a processor, a plurality of documents related to business requirements. The embodiment may also include parsing the received documents to extract business concepts based on sentence analysis utilizing an NLP technology. The embodiment may further include generating domain models based on the extracted business concepts. The embodiment may also include clustering the generated domain models into specific domains.
-
公开(公告)号：US20230103149A1
公开(公告)日：2023-03-30
申请号：US17449652
申请日：2021-09-30
Applicant: International Business Machines Corporation
Inventor： Chao Yu , Kun Yan Yin , Ming Jin Chen , Teng Sun , Guang Qing Zhong
IPC: G06N3/04
Abstract: An approach is provided for adaptively compressing a deep learning model. An original deep learning model for different Internet of Things (IoT) devices is determined. Device information is collected from the IoT devices. Based on the device information, multiple recommendation engines are selected from a set of recommendation engines. Compression factor combinations are determined by using the multiple recommendation engines. Compression ratios and model accuracies for the compression factor combinations are determined. Based on the compression ratios and the model accuracies, an optimal compression factor combination is selected from the compression factor combinations. A compressed deep learning model is generated by compressing the original deep learning model by using the optimal compression factor. The compressed deep learning model is deployed to the IoT devices.
-
公开(公告)号：US11615618B2
公开(公告)日：2023-03-28
申请号：US17225165
申请日：2021-04-08
Applicant: International Business Machines Corporation
Inventor： Kun Yan Yin , Chao Yu , Ming Jin Chen , Teng Sun , Hong Bing Zhang
Abstract: A computer-implemented method for annotating images is disclosed. The computer-implemented method includes generating a saliency map corresponding to an input image, wherein the input image is an image that requires annotation, generating a behavior saliency map, wherein the behavior saliency map is a saliency map formed from an average of a plurality of objects contained within respective bounding boxes of a plurality of sample images, generating a historical saliency map, wherein the historical saliency map is a saliency map formed from an average of a plurality of tagged objects in the plurality of sample images, fusing the saliency map corresponding to the input image, the behavior saliency map, and the historical saliency map to form a fused saliency map, and generating, based on the fused saliency map, a bounding box around an object in the input image.
-
公开(公告)号：US20220327312A1
公开(公告)日：2022-10-13
申请号：US17225165
申请日：2021-04-08
Applicant: International Business Machines Corporation
Inventor： Kun Yan Yin , Chao Yu , Ming Jin Chen , Teng Sun , Hong Bing Zhang
Abstract: A computer-implemented method for annotating images is disclosed. The computer-implemented method includes generating a saliency map corresponding to an input image, wherein the input image is an image that requires annotation, generating a behavior saliency map, wherein the behavior saliency map is a saliency map formed from an average of a plurality of objects contained within respective bounding boxes of a plurality of sample images, generating a historical saliency map, wherein the historical saliency map is a saliency map formed from an average of a plurality of tagged objects in the plurality of sample images, fusing the saliency map corresponding to the input image, the behavior saliency map, and the historical saliency map to form a fused saliency map, and generating, based on the fused saliency map, a bounding box around an object in the input image.
-
Patent Agency Ranking
公开(公告)号：US20250029405A1
公开(公告)日：2025-01-23
申请号：US18353392
申请日：2023-07-17
Applicant: International Business Machines Corporation
Inventor： Yu Zui You , Teng Sun , Yi Ming Wang , Hai Bo Zou
Abstract: Vehicle license plate auditing is provided. Distance and time traveling information corresponding to a vehicle is extracted from a data network analysis graph model of spatiotemporal traffic flow data based on a license plate number of the vehicle in response to removing irrelevant nodes from the data network analysis graph model. A distance and time mutually exclusive node relationship is determined within the data network analysis graph model of spatiotemporal traffic flow data based on the distance and time traveling information corresponding to the vehicle extracted from the data network analysis graph model of spatiotemporal traffic flow data. An anomalous license plate node is determined in the data network analysis graph model of spatiotemporal traffic flow data based on the distance and time mutually exclusive node relationship. A set of action steps is performed regarding an anomalous license plate number corresponding to the anomalous license plate node.
-
公开(公告)号：US20220351020A1
公开(公告)日：2022-11-03
申请号：US17245541
申请日：2021-04-30
Applicant: International Business Machines Corporation
Inventor： Kun Yan Yin , Chao Yu , Ming Jin Chen , Teng Sun , Xiao Ye Li
IPC: G06N3/04
Abstract: In an approach to deploying parallelizable deep learning models by adapting to the computing devices, a deep learning model is split into a plurality of slices, where each slice can exchange data with related slices. Virtual models are created from the plurality of slices, where the virtual models are based on capabilities of a plurality of devices on which the one or more virtual models are to be deployed, and further where each virtual model contains each slice of the plurality of slices. The one or more virtual models are stored in a cache. Responsive to determining that the deep learning model is to be deployed on one or more devices, a candidate model is selected from the virtual models in the cache, where the selection is based on information from a device monitor about the devices.
-
公开(公告)号：US20180336018A1
公开(公告)日：2018-11-22
申请号：US15596077
申请日：2017-05-16
Applicant: International Business Machines Corporation
Inventor： Liang KS Lu , Teng Sun , Zhong Shi Wang , Zhe LI Yu
IPC: G06F9/44
Abstract: The present invention is a system and method for template extraction based on source code similarity. The system receives source code and groups the class files into classes based on naming rules and inheritance hierarchy. Features are parsed for each class and encoded a float value. The classes are clustered based on similarities of the features. A similarity value is calculated for the classes in a cluster and potential candidate classes are selected based on the similarity value or inheritance hierarchy. A feature subset is selected across all candidate classes and differences in the features in the subset are determined. The features are then decoded and the differences are parameterized to generate a template. A variable definition file is created to cross-reference features and variables. Source code can then be generated using the template and the variable definition file.
-
公开(公告)号：US20230076923A1
公开(公告)日：2023-03-09
申请号：US17467516
申请日：2021-09-07
Applicant: International Business Machines Corporation
Inventor： Teng Sun , Tong Liu , Si Tong Zhao , XueLiang Zhao , Frank Feng , Yu Zui WY You , Zhong Fang Yuan
IPC: G06F16/2453 , G06F16/901 , G06F16/242 , G06F16/21 , G06N20/00
Abstract: In order to perform a semantic search based on a graph database, sets of nodes are selected from a plurality of nodes in a graph database. A set of nodes semantically matches a keyword in a natural language query. At least one target node is identified in the sets of nodes. A path is selected from candidate paths based on similarities between the candidate paths and a plurality of paths in the graph database. A graph query for retrieving information from the graph database is generated based on the selected path and the query target.
-
公开(公告)号：US10296307B2
公开(公告)日：2019-05-21
申请号：US15596077
申请日：2017-05-16
Applicant: International Business Machines Corporation
Inventor： Liang K S Lu , Teng Sun , Zhong Shi Wang , Zhe Li Yu
Abstract: The present invention is a system and method for template extraction based on source code similarity. The system receives source code and groups the class files into classes based on naming rules and inheritance hierarchy. Features are parsed for each class and encoded a float value. The classes are clustered based on similarities of the features. A similarity value is calculated for the classes in a cluster and potential candidate classes are selected based on the similarity value or inheritance hierarchy. A feature subset is selected across all candidate classes and differences in the features in the subset are determined. The features are then decoded and the differences are parameterized to generate a template. A variable definition file is created to cross-reference features and variables. Source code can then be generated using the template and the variable definition file.
-
-
-
-
-
-
-
-
-
Search Results
10
records
(took 0.013 seconds)
