Memorization with small neural networks

We present simple constructions of memorizing neural network with optimal number of neurons. In particular we describe how this can be achieved in the NTK (neural tangent kernel) regime.
Based on ``Network size and weights size for memorization with two-layers neural networks" by S.B., R. Eldan, Y.T. Lee and D. Mikulincer. arxiv.org/abs/2006.02855

[@aaryanbhagat4852]

Nice explanation!