On the Fundamental Limits of Coded Data Shuffling for Distributed Machine Learning | IEEE Journals & Magazine | IEEE Xplore