Skip to yearly menu bar Skip to main content


Poster

Constructive Universal High-Dimensional Distribution Generation through Deep ReLU Networks

Dmytro Perekrestenko · Stephan Müller · Helmut Bölcskei

Virtual

Keywords: [ Deep Learning Theory ] [ Other ] [ Deep Learning - Theory ]


Abstract: We present an explicit deep neural network construction that transforms uniformly distributed one-dimensional noise into an arbitrarily close approximation of any two-dimensional Lipschitz-continuous target distribution. The key ingredient of our design is a generalization of the "space-filling" property of sawtooth functions discovered in (Bailey & Telgarsky, 2018). We elicit the importance of depth - in our neural network construction - in driving the Wasserstein distance between the target distribution and the approximation realized by the network to zero. An extension to output distributions of arbitrary dimension is outlined. Finally, we show that the proposed construction does not incur a cost - in terms of error measured in Wasserstein-distance - relative to generating $d$-dimensional target distributions from $d$ independent random variables.

Chat is not available.