Fetching the paper…

Expressivity and Approximation Properties of Deep Neural Networks with ReLU$^k$ Activation · Around