Shallow Neural Networks Learn Low-Degree Spherical Polynomials with Learnable Channel Attention

Open in new window