公开(公告)号：US20200151563A1

公开(公告)日：2020-05-14

申请号：US16675596

申请日：2019-11-06

Applicant: NEC Laboratories America, Inc.

Inventor： Bo Zong ,  Jingchao Ni ,  Haifeng Chen ,  Cheng Zheng

IPC: G06N3/08 ,  G06N3/04

Abstract: A method for employing a supervised graph sparsification (SGS) network to use feedback from subsequent graph learning tasks to guide graph sparsification is presented. The method includes, in a training phase, generating sparsified subgraphs by edge sampling from input training graphs following a learned distribution, feeding the sparsified subgraphs to a prediction/classification component, collecting a predication/classification error, and updating parameters of the learned distribution based on a gradient derived from the predication/classification error. The method further includes, in a testing phase, generating sparsified subgraphs by edge sampling from input testing graphs following the learned distribution, feeding the sparsified subgraphs to the prediction/classification component, and outputting prediction/classification results to a visualization device.

公开(公告)号：US11645540B2

公开(公告)日：2023-05-09

申请号：US16936600

申请日：2020-07-23

Applicant: NEC Laboratories America, Inc.

Inventor： Bo Zong ,  Cheng Zheng ,  Haifeng Chen

IPC: G06V10/774 ,  G06K9/62 ,  G06F11/07 ,  G06N3/04 ,  G06V30/196

CPC classification number: G06V10/7747 ,  G06F11/0766 ,  G06K9/6223 ,  G06K9/6257 ,  G06K9/6276 ,  G06N3/0454 ,  G06V30/1988

Abstract: A method for employing a differentiable ranking based graph sparsification (DRGS) network to use supervision signals from downstream tasks to guide graph sparsification is presented. The method includes, in a training phase, generating node representations by neighborhood aggregation operators, generating sparsified subgraphs by top-k neighbor sampling from a learned neighborhood ranking distribution, feeding the sparsified subgraphs to a task, generating a prediction, and collecting a prediction error to update parameters in the generating and feeding steps to minimize an error, and, in a testing phase, generating node representations by neighborhood aggregation operators related to testing data, generating sparsified subgraphs by top-k neighbor sampling from a learned neighborhood ranking distribution related to the testing data, feeding the sparsified subgraphs related to the testing data to a task, and outputting prediction results to a visualization device.

公开(公告)号：US20210049414A1

公开(公告)日：2021-02-18

申请号：US16936600

申请日：2020-07-23

Applicant: NEC Laboratories America, Inc.

Inventor： Bo Zong ,  Cheng Zheng ,  Haifeng Chen

IPC: G06K9/62 ,  G06K9/68 ,  G06N3/04 ,  G06F11/07

Abstract: A method for employing a differentiable ranking based graph sparsification (DRGS) network to use supervision signals from downstream tasks to guide graph sparsification is presented. The method includes, in a training phase, generating node representations by neighborhood aggregation operators, generating sparsified subgraphs by top-k neighbor sampling from a learned neighborhood ranking distribution, feeding the sparsified subgraphs to a task, generating a prediction, and collecting a prediction error to update parameters in the generating and feeding steps to minimize an error, and, in a testing phase, generating node representations by neighborhood aggregation operators related to testing data, generating sparsified subgraphs by top-k neighbor sampling from a learned neighborhood ranking distribution related to the testing data, feeding the sparsified subgraphs related to the testing data to a task, and outputting prediction results to a visualization device.

公开(公告)号：US11610114B2

公开(公告)日：2023-03-21

申请号：US16675596

申请日：2019-11-06

Applicant: NEC Laboratories America, Inc.

Inventor： Bo Zong ,  Jingchao Ni ,  Haifeng Chen ,  Cheng Zheng

IPC: G06N3/08 ,  G06N3/04

Abstract: A method for employing a supervised graph sparsification (SGS) network to use feedback from subsequent graph learning tasks to guide graph sparsification is presented. The method includes, in a training phase, generating sparsified subgraphs by edge sampling from input training graphs following a learned distribution, feeding the sparsified subgraphs to a prediction/classification component, collecting a predication/classification error, and updating parameters of the learned distribution based on a gradient derived from the predication/classification error. The method further includes, in a testing phase, generating sparsified subgraphs by edge sampling from input testing graphs following the learned distribution, feeding the sparsified subgraphs to the prediction/classification component, and outputting prediction/classification results to a visualization device.