G RAD -M ATCH: Gradient Matching based Data Subset Selection for Efficient
             Deep Model Training

  Krishnateja Killamsetty 1 Durga Sivasubramanian 2 Ganesh Ramakrishnan 2 Abir De 2 Rishabh Iyer 1

             Abstract               loss in test accuracy?
    The great success of modern machine learning       Data subset selection enables efficient learning at multiple
    models on large datasets is contingent on exten-     levels. First, by using a subset of a large dataset, we can ...