Feasibility-based Fixed Point Networks
arXiv:2104.14090
Abstract
Inverse problems consist of recovering a signal from a collection of noisy measurements. These problems can often be cast as feasibility problems; however, additional regularization is typically necessary to ensure accurate and stable recovery with respect to data perturbations. Hand-chosen analytic regularization can yield desirable theoretical guarantees, but such approaches have limited effectiveness recovering signals due to their inability to leverage large amounts of available data. To this end, this work fuses data-driven regularization and convex feasibility in a theoretically sound manner. This is accomplished using feasibility-based fixed point networks (F-FPNs). Each F-FPN defines a collection of nonexpansive operators, each of which is the composition of a projection-based operator and a data-driven regularization operator. Fixed point iteration is used to compute fixed points of these operators, and weights of the operators are tuned so that the fixed points closely represent available data. Numerical examples demonstrate performance increases by F-FPNs when compared to standard TV-based recovery methods for CT reconstruction and a comparable neural network based on algorithm unrolling.
References in corpus (23)
- PyTorch: An Imperative Style, High-Performance Deep Learning Library
- Deep Convolutional Neural Network for Inverse Problems in Imaging
- Trainable Nonlinear Reaction Diffusion: A Flexible Framework for Fast and Effective Image Restoration
- Deep Equilibrium Models
- Trainable ISTA for Sparse Signal Recovery
- Learned D-AMP: Principled Neural Network based Compressive Image Recovery
- Perturbation Resilience and Superiorization of Iterative Algorithms
- Theoretical Linear Convergence of Unfolded ISTA and its Practical Weights and Thresholds
- Learned Convolutional Sparse Coding
- Unrolled Optimization with Deep Priors
- Neural Proximal Gradient Descent for Compressive Imaging
- Recurrent Inference Machines for Solving Inverse Problems
- Learning to Optimize: A Primer and A Benchmark
- Superiorization: An optimization heuristic for medical physics
- Convergence and Perturbation Resilience of Dynamic String-Averaging Projection Methods
- Projection Methods: An Annotated Bibliography of Books and Reviews
- Differentiable Linearized ADMM
- Universal Approximation Power of Deep Residual Neural Networks via Nonlinear Control Theory
- Learned convex regularizers for inverse problems
- A Multiscale Method for Model Order Reduction in PDE Parameter Estimation
- PNKH-B: A Projected Newton-Krylov Method for Large-Scale Bound-Constrained Optimization
- Degrees of Freedom Analysis of Unrolled Neural Networks
- Superiorization and Perturbation Resilience of Algorithms: A Continuously Updated Bibliography