Optimal bump functions for shallow ReLU networks: Weight decay, depth separation and the curse of dimensionality

09/02/2022
by   Stephan Wojtowytsch, et al.
0

In this note, we study how neural networks with a single hidden layer and ReLU activation interpolate data drawn from a radially symmetric distribution with target labels 1 at the origin and 0 outside the unit ball, if no labels are known inside the unit ball. With weight decay regularization and in the infinite neuron, infinite data limit, we prove that a unique radially symmetric minimizer exists, whose weight decay regularizer and Lipschitz constant grow as d and √(d) respectively. We furthermore show that the weight decay regularizer grows exponentially in d if the label 1 is imposed on a ball of radius ε rather than just at the origin. By comparison, a neural networks with two hidden layers can approximate the target function without encountering the curse of dimensionality.

READ FULL TEXT

Please sign up or login with your details

Forgot password? Click here to reset