Search Patents
  • Recommending exemplars of an unlabeled data-set
    Patent number: 10838968
    Abstract: Embodiments for recommending exemplars of a data-set by a processor. A selected number of exemplars may be labeled from one or more classes in a data-set. One or more class exemplars for each of the one or more classes in the data-set may be recommended according to similarities between the selected number of labeled exemplars and remaining data of the data-set.
    Type: Grant
    Filed: January 10, 2018
    Date of Patent: November 17, 2020
    Assignee: INTERNATIONAL BUSINESS MACHINES CORPORATION
    Inventors: Shrihari Vasudevan, Joydeep Mondal, Richard H. Zhou, Michael Peran, Michael W. Ticknor, Daniel Augenstern