Image Reconstruction via Deep Image Prior Subspaces
This work addresses the deployability issues of unsupervised deep learning for image reconstruction, offering a solution for scenarios with limited training data, though it is incremental in improving existing DIP methods.
The paper tackled the problem of deep image prior (DIP) overfitting and unstable convergence in unsupervised image reconstruction by restricting optimization to a sparse linear subspace of parameters, which reduced noise fitting and enabled stable second-order methods. Experiments across various image restoration and tomographic tasks showed improved optimization stability and reconstruction fidelity trade-off.
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 training data. Unsupervised learning methods, such as the 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.