Whiteness-based bilevel learning of regularization parameters in imaging
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866909134617575424 |
|---|---|
| author | Santambrogio, Carlo Pragliola, Monica Lanza, Alessandro Donatelli, Marco Calatroni, Luca |
| author_facet | Santambrogio, Carlo Pragliola, Monica Lanza, Alessandro Donatelli, Marco Calatroni, Luca |
| contents | We consider an unsupervised bilevel optimization strategy for learning regularization parameters in the context of imaging inverse problems in the presence of additive white Gaussian noise. Compared to supervised and semi-supervised metrics relying either on the prior knowledge of reference data and/or on some (partial) knowledge on the noise statistics, the proposed approach optimizes the whiteness of the residual between the observed data and the observation model with no need of ground-truth data.We validate the approach on standard Total Variation-regularized image deconvolution problems which show that the proposed quality metric provides estimates close to the mean-square error oracle and to discrepancy-based principles. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2403_07026 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Whiteness-based bilevel learning of regularization parameters in imaging Santambrogio, Carlo Pragliola, Monica Lanza, Alessandro Donatelli, Marco Calatroni, Luca Optimization and Control Machine Learning Image and Video Processing We consider an unsupervised bilevel optimization strategy for learning regularization parameters in the context of imaging inverse problems in the presence of additive white Gaussian noise. Compared to supervised and semi-supervised metrics relying either on the prior knowledge of reference data and/or on some (partial) knowledge on the noise statistics, the proposed approach optimizes the whiteness of the residual between the observed data and the observation model with no need of ground-truth data.We validate the approach on standard Total Variation-regularized image deconvolution problems which show that the proposed quality metric provides estimates close to the mean-square error oracle and to discrepancy-based principles. |
| title | Whiteness-based bilevel learning of regularization parameters in imaging |
| topic | Optimization and Control Machine Learning Image and Video Processing |
| url | https://arxiv.org/abs/2403.07026 |