Scaling Distributed Machine Learning with the Parameter Server

  Abstract