公开(公告)号：US10303801B2

公开(公告)日：2019-05-28

申请号：US15390239

申请日：2016-12-23

Applicant: INTERNATIONAL BUSINESS MACHINES CORPORATION

Inventor： Matthew L. Hill ,  John R. Kender ,  Apostol I. Natsev ,  John R. Smith ,  Lexing Xie

IPC: G06F7/00 ,  G06F17/30

Abstract: A system and method for analyzing visual memes includes identifying visual memes associated with at least one topic in a data source. The visual memes propagated over time are tracked to extract information associated with identified visual memes. The information associated with the visual memes is analyzed to determine at least one of generation, propagation, and use of the identified memes.

公开(公告)号：US20200082210A1

公开(公告)日：2020-03-12

申请号：US16125153

申请日：2018-09-07

Applicant: International Business Machines Corporation

Inventor： Patrick Watson ,  Bishwaranjan Bhattacharjee ,  Siyu Huo ,  Noel C. Codella ,  Brian M. Belgodere ,  Parijat Dube ,  Michael R. Glass ,  John R. Kender ,  Matthew L. Hill

IPC: G06K9/62 ,  G06F17/30

Abstract: A computer-implemented method for data labeling is provided. The computer-implemented method assigns pseudo-labels to unlabeled examples of data using a similarity metric on an embedding space to produce pseudo-labeled examples. A curriculum learning model is trained using the pseudo-labeled examples. The curriculum learning model trained with the pseudo-labeled examples is employed in in a fine-tuning task to enhance classification accuracy of the data.

公开(公告)号：US11151410B2

公开(公告)日：2021-10-19

申请号：US16125153

申请日：2018-09-07

Applicant: International Business Machines Corporation

Inventor： Patrick Watson ,  Bishwaranjan Bhattacharjee ,  Siyu Huo ,  Noel C. Codella ,  Brian M. Belgodere ,  Parijat Dube ,  Michael R. Glass ,  John R. Kender ,  Matthew L. Hill

IPC: G06K9/62 ,  G06F16/51

Abstract: A computer-implemented method for data labeling is provided. The computer-implemented method assigns pseudo-labels to unlabeled examples of data using a similarity metric on an embedding space to produce pseudo-labeled examples. A curriculum learning model is trained using the pseudo-labeled examples. The curriculum learning model trained with the pseudo-labeled examples is employed in in a fine-tuning task to enhance classification accuracy of the data.