Abstract
Abstract
The analytic deep prior (ADP) approach was recently introduced for the theoretical analysis of deep image prior (DIP) methods with special network architectures. In this paper, we prove that ADP is in fact equivalent to classical variational Ivanov methods for solving ill-posed inverse problems. Besides, we propose a new variant which incorporates the strategy of early stopping into the ADP model. For both variants, we show how classical regularization properties (existence, stability, convergence) can be obtained under common assumptions.
Subject
Applied Mathematics,Computer Science Applications,Mathematical Physics,Signal Processing,Theoretical Computer Science
Reference38 articles.
1. Learning the optimal Tikhonov regularizer for inverse problems;Alberti,2021
2. Solving inverse problems using data-driven models;Arridge;Acta Numer.,2019
3. Computed tomography reconstruction using deep image prior and learned reconstruction methods;Baguer;Inverse Problems,2020
4. Regularization and complexity control in feed-forward networks;Bishop,1995
5. A double regularization approach for inverse problems with noisy data and inexact operator;Bleyer;Inverse Problems,2013
Cited by
2 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献