Despite the exciting progress in the field and the abundance of new methods proposed, how these algorithms compare against each other remains unclear.

Second-order information, in the form of Hessian-or Inverse-Hessian-vector products, is a fundamental tool for solving optimization problems.

We also devise a reinforcement learning algorithm for SICMDPs that we call SI-CRL.

We develop the first polynomial-time algorithms that return a near-optimal solution with high probability.

Multi-Task Learning (MTL) is a powerful learning paradigm to improve generalization performance via knowledge sharing.

This shows that local minibatching and device sampling are both beneficial for linearly speeding up communication. A few observations are in order.