Training Quantized Neural Networks to Global Optimality via Semidefinite
                Programming

                       Burak Bartan 1 Mert Pilanci 1

            Abstract                dimensions using semidefinite programming (Bartan & Pi-
   Neural networks (NNs) have been extremely suc-       lanci, 2021). In this work, we take a similar convex duality
   cessful across many tasks in machine learning.       approach that involves semidefinite programming. However,
   Quantization of NN ...