Optimization-Inspired Learning with Architecture Augmentations and Control Mechanisms for Low-Level Vision
This work provides a unified framework and theoretical guarantees for combining learnable modules with numerical optimization in low-level vision, which is an incremental improvement for researchers in this specific domain.
This paper proposes a unified optimization-inspired learning framework, GDC, that aggregates Generative, Discriminative, and Corrective principles to solve low-level vision tasks. It constructs three propagative modules based on different descent direction formulations of a general energy minimization model and designs two control mechanisms with theoretical guarantees for stable propagation and convergence.
In recent years, there has been a growing interest in combining learnable modules with numerical optimization to solve low-level vision tasks. However, most existing approaches focus on designing specialized schemes to generate image/feature propagation. There is a lack of unified consideration to construct propagative modules, provide theoretical analysis tools, and design effective learning mechanisms. To mitigate the above issues, this paper proposes a unified optimization-inspired learning framework to aggregate Generative, Discriminative, and Corrective (GDC for short) principles with strong generalization for diverse optimization models. Specifically, by introducing a general energy minimization model and formulating its descent direction from different viewpoints (i.e., in a generative manner, based on the discriminative metric and with optimality-based correction), we construct three propagative modules to effectively solve the optimization models with flexible combinations. We design two control mechanisms that provide the non-trivial theoretical guarantees for both fully- and partially-defined optimization formulations. Under the support of theoretical guarantees, we can introduce diverse architecture augmentation strategies such as normalization and search to ensure stable propagation with convergence and seamlessly integrate the suitable modules into the propagation respectively. Extensive experiments across varied low-level vision tasks validate the efficacy and adaptability of GDC. The codes are available at https://github.com/LiuZhu-CV/GDC-OptimizationLearning