Rates of Approximation by ReLU Shallow Neural Networks

Tong Mao, Ding-Xuan Zhou

Neural networks activated by the rectified linear unit (ReLU) play a central role in the recent development of deep learning. The topic of approximating functions from Hölder space…