18
2

Deep Neural Networks: Multi-Classification and Universal Approximation

Abstract

We demonstrate that a ReLU deep neural network with a width of 22 and a depth of 2N+4M12N+4M-1 layers can achieve finite sample memorization for any dataset comprising NN elements in Rd\mathbb{R}^d, where d1,d\ge1, and MM classes, thereby ensuring accurate classification. By modeling the neural network as a time-discrete nonlinear dynamical system, we interpret the memorization property as a problem of simultaneous or ensemble controllability. This problem is addressed by constructing the network parameters inductively and explicitly, bypassing the need for training or solving any optimization problem. Additionally, we establish that such a network can achieve universal approximation in Lp(Ω;R+)L^p(\Omega;\mathbb{R}_+), where Ω\Omega is a bounded subset of Rd\mathbb{R}^d and p[1,)p\in[1,\infty), using a ReLU deep neural network with a width of d+1d+1. We also provide depth estimates for approximating W1,pW^{1,p} functions and width estimates for approximating Lp(Ω;Rm)L^p(\Omega;\mathbb{R}^m) for m1m\geq1. Our proofs are constructive, offering explicit values for the biases and weights involved.

View on arXiv
Comments on this paper