Recovery Guarantees of Unsupervised Neural Networks for Inverse Problems trained with Gradient Descent

Nathan Buskulic , Jalal Fadili (1, 2, 3, 4) , Yvain Quéau (1)

1 Equipe Image - Laboratoire GREYC - UMR6072
2 NU - Normandie Université
3 ENSICAEN - École Nationale Supérieure d'Ingénieurs de Caen
4 GREYC - Groupe de Recherche en Informatique, Image et Instrumentation de Caen

Advanced machine learning methods, and more prominently neural networks, have become standard to solve inverse problems over the last years. However, the theoretical recovery guarantees of such methods are still scarce and difficult to achieve. Only recently did unsupervised methods such as Deep Image Prior (DIP) get equipped with convergence and recovery guarantees for generic loss functions when trained through gradient flow with an appropriate initialization. In this paper, we extend these results by proving that these guarantees hold true when using gradient descent with an appropriately chosen step-size/learning rate. We also show that the discretization only affects the overparametrization bound for a two-layer DIP network by a constant and thus that the different guarantees found for the gradient flow will hold for gradient descent.

Mots clés: Inverse problems; Deep Image/Inverse Prior; Overparametrization; Gradient descent; Unsupervised learning

Domaines: Apprentissage [cs.LG]

Nathan Buskulic, Jalal Fadili, Yvain Quéau. Recovery Guarantees of Unsupervised Neural Networks for Inverse Problems trained with Gradient Descent. 2024. ⟨hal-04496435⟩