Fetching the paper…

High-Order Approximation Rates for Shallow Neural Networks with Cosine and ReLU$^k$ Activation Functions · Around