English

Parseval Proximal Neural Networks

Numerical Analysis 2020-11-30 v2 Numerical Analysis

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 H\mathcal H and K\mathcal K be real Hilbert spaces, bKb\in\mathcal K and TB(H,K)T\in\mathcal{B}(\mathcal H,\mathcal K) have closed range and Moore-Penrose inverse TT^\dagger. Based on the well-known characterization of proximity operators by Moreau, we prove that for any proximity operator Prox ⁣:KK\text{Prox}\colon\mathcal K\to\mathcal K the operator TProx(T+b)T^\dagger\,\text{Prox} (T\cdot +b) is a proximity operator on H\mathcal H equipped with a suitable norm. In particular, it follows for the frequently applied soft shrinkage operator Prox=Sλ ⁣:22\text{Prox} = S_{\lambda}\colon\ell_2 \rightarrow\ell_2 and any frame analysis operator T ⁣:H2T\colon\mathcal H\to\ell_2 that the frame shrinkage operator TSλTT^\dagger\, S_\lambda\,T is a proximity operator on a suitable Hilbert space. The concatenation of proximity operators on Rd\mathbb R^d 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.

Keywords

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

R2 v1 2026-06-23T12:53:51.094Z