Image Reconstruction via Deep Image Prior Subspaces

Published: 18 Feb 2024, Last Modified: 18 Feb 2024Accepted by TMLREveryoneRevisionsBibTeX
Abstract: Deep learning has been widely used for solving image reconstruction tasks but its deployability has been held back due to the shortage of high-quality paired training data. Unsupervised learning methods, e.g., deep image prior (DIP), naturally fill this gap, but bring a host of new issues: the susceptibility to overfitting due to a lack of robust early stopping strategies and unstable convergence. We present a novel approach to tackle these issues by restricting DIP optimisation to a sparse linear subspace of its parameters, employing a synergy of dimensionality reduction techniques and second order optimisation methods. The low-dimensionality of the subspace reduces DIP's tendency to fit noise and allows the use of stable second order optimisation methods, e.g., natural gradient descent or L-BFGS. Experiments across both image restoration and tomographic tasks of different geometry and ill-posedness show that second order optimisation within a low-dimensional subspace is favourable in terms of optimisation stability to reconstruction fidelity trade-off.
Submission Length: Regular submission (no more than 12 pages of main content)
Supplementary Material: pdf
Assigned Action Editor: ~Jeremias_Sulam1
License: Creative Commons Attribution 4.0 International (CC BY 4.0)
Submission Number: 1604