Parseval Proximal Neural Networks
Abstract
The aim of this paper is twofold. First, we show that a certain concatenation of a proximity operator with an affine operator is again a proximity operator on a suitable Hilbert space. Second, we use our findings to establish so-called proximal neural networks (PNNs) and stable tight frame proximal neural networks. Let and be real Hilbert spaces, and have closed range and Moore-Penrose inverse . Based on the well-known characterization of proximity operators by Moreau, we prove that for any proximity operator the operator is a proximity operator on equipped with a suitable norm. In particular, it follows for the frequently applied soft shrinkage operator and any frame analysis operator that the frame shrinkage operator is a proximity operator on a suitable Hilbert space. The concatenation of proximity operators on equipped with different norms establishes a PNN. If the network arises from tight frame analysis or synthesis operators, then it forms an averaged operator. Hence, it has Lipschitz constant 1 and belongs to the class of so-called Lipschitz networks, which were recently applied to defend against adversarial attacks. Moreover, due to its averaging property, PNNs can be used within so-called Plug-and-Play algorithms with convergence guarantee. In case of Parseval frames, we call the networks Parseval proximal neural networks (PPNNs). Then, the involved linear operators are in a Stiefel manifold and corresponding minimization methods can be applied for training. Finally, some proof-of-the concept examples demonstrate the performance of PPNNs.
Cite
@article{arxiv.1912.10480,
title = {Parseval Proximal Neural Networks},
author = {Marzieh Hasannasab and Johannes Hertrich and Sebastian Neumayer and Gerlind Plonka and Simon Setzer and Gabriele Steidl},
journal= {arXiv preprint arXiv:1912.10480},
year = {2020}
}
Comments
arXiv admin note: text overlap with arXiv:1910.02843