A Theory of High Dimensional Regression with Arbitrary Correlations
between Input Features and Target Functions: Sample Complexity, Multiple
      Descent Curves and a Hierarchy of Phase Transitions

                       Gabriel C. Mel 1 Surya Ganguli 2

             Abstract                understanding the ability of not only deep networks, but also
                                many other machine learning (ML) methods, to successfully
   The performance of neural networks depends on      ...