When does gradient descent with logistic loss interpolate using deep networks with smoothed ReLU activations?
Proceedings of Thirty Fourth Conference on Learning Theory, PMLR 134:927-1027, 2021.
We establish conditions under which gradient descent applied to fixed-width deep networks drives the logistic loss to zero, and prove bounds on the rate of convergence. Our analysis applies for smoothed approximations to the ReLU, such as Swish and the Huberized ReLU, proposed in previous applied work. We provide two sufficient conditions for convergence. The first is simply a bound on the loss at initialization. The second is a data separation condition used in prior analyses.