Technology Category
Computer Vision: Learning & Optimization for CVSearch and Optimization: Learning to SearchConstraint Satisfaction and Optimization: Constraint Learning and Acquisition
LAMA-Net: A Convergent Network Architecture for Dual-Domain Reconstruction
🤖 AI Summary
To address the challenges of fusing image-domain and measurement-domain features in image reconstruction and the lack of convergence guarantees in learnable optimization algorithms, this paper proposes iLAMA-Net: a learnable variational model with rigorous convergence guarantees. Methodologically, it introduces a dual-domain joint optimization framework based on a Learnable Initial-value Proximal Alternating Minimization Algorithm (LAMA), where deep networks are embedded via residual learning to enable synergistic reconstruction leveraging complementary information from both domains. Theoretically, this work provides the first convergence proof for learnable alternating minimization, establishing that all accumulation points are Clarke stationary points. Technically, the learnable initialization module significantly enhances reconstruction quality. Evaluated on sparse-view CT benchmark datasets, iLAMA-Net consistently outperforms state-of-the-art methods in reconstruction accuracy, stability, and robustness.
Technology Category
Application Category
📝 Abstract
We propose a learnable variational model that learns the features and leverages complementary information from both image and measurement domains for image reconstruction. In particular, we introduce a learned alternating minimization algorithm (LAMA) from our prior work, which tackles two-block nonconvex and nonsmooth optimization problems by incorporating a residual learning architecture in a proximal alternating framework. In this work, our goal is to provide a complete and rigorous convergence proof of LAMA and show that all accumulation points of a specified subsequence of LAMA must be Clarke stationary points of the problem. LAMA directly yields a highly interpretable neural network architecture called LAMA-Net. Notably, in addition to the results shown in our prior work, we demonstrate that the convergence property of LAMA yields outstanding stability and robustness of LAMA-Net in this work. We also show that the performance of LAMA-Net can be further improved by integrating a properly designed network that generates suitable initials, which we call iLAMA-Net. To evaluate LAMA-Net/iLAMA-Net, we conduct several experiments and compare them with several state-of-the-art methods on popular benchmark datasets for Sparse-View Computed Tomography.
Problem
Research questions and friction points this paper is trying to address.
Dual-domain image reconstruction using learnable variational model
Convergence proof for learned alternating minimization algorithm (LAMA)
Improving stability and robustness in neural network architecture
Innovation
Methods, ideas, or system contributions that make the work stand out.
Learnable variational model for dual-domain reconstruction
Learned alternating minimization algorithm with convergence proof
Interpretable neural network architecture with stability
🔎 Similar Papers
No similar papers found.
💼 Related Jobs
No related jobs found.
