eduzhai > Applied Sciences > Engineering >

openXDATA A Tool for Multi-Target Data Generation and Missing Label Completion

  • king
  • (0) Download
  • 20210506
  • Save

... pages left unread,continue reading

Document pages: 5 pages

Abstract: A common problem in machine learning is to deal with datasets with disjointlabel spaces and missing labels. In this work, we introduce the openXDATA toolthat completes the missing labels in partially labelled or unlabelled datasetsin order to generate multi-target data with labels in the joint label space ofthe datasets. To this end, we designed and implemented the cross-data labelcompletion (CDLC) algorithm that uses a multi-task shared-hidden-layer DNN toiteratively complete the sparse label matrix of the instances from thedifferent datasets. We apply the new tool to estimate labels across fouremotion datasets: one labeled with discrete emotion categories (e.g., happy,sad, angry), one labeled with continuous values along arousal and valencedimensions, one with both kinds of labels, and one unlabeled. Testing withdrop-out of true labels, we show the ability to estimate both categories andcontinuous labels for all of the datasets, at rates that approached the groundtruth values. openXDATA is available under the GNU General Public License fromthis https URL.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×