AI Networking Cookbook: Practical recipes for AI-assisted network automation and development
TL;DR
A model-based deep denoising framework that integrates three tightly coupled components, and proves that every limit point of the L-ADMM sequence is a stationary point of the nonconvex objective, and that the iterate sequence itself converges globally.
Experimental results on both grayscale and color image restoration demonstrate that the proposed method consistently outperforms representative model-based approaches while achieving performance comparable to state-of-the-art DEQ models based on implicit regularization, despite requiring substantially fewer trainable parameters.
GSDSplit+ is introduced, a splitting method built upon PnPSplit+, in which the generic denoising block is replaced by a relaxed Gradient Step Denoiser, which retains explicit Poisson fidelity updates.
This paper proposes an ADMM-inspired unrolled plug-and-play solver for Poisson inverse problems that decouples a closed-form data-consistency update from a parameter-efficient prior, implemented as a lightweight decoder operating on frozen CLIP RN50 dense multi-scale features.
This paper proposes a novel learned spherical alternating direction method of multipliers (LSADMMs) for effective Rician noise removal under spherical constraints, and incorporates a noise-level estimation prefix that provides adaptive guidance across noise levels.
Abstract. We describe a practical framework for data-driven regularization in image reconstruction. It combines model expressivity with the guarantees of variational methods. The approach is based on a weakly convex ridge regularizer, defined as the composition of a convolutional filter bank and pointwise potentials constrained to be weakly convex. These potentials are implemented as learnable splines with a strict control of their weak-convexity modulus. The resulting denoisers outperform classic convex regularization techniques as well as competitive benchmarks such as BM3D, while they still correspond to the minimization of a convex energy. The learned regularizers further extend to general inverse problems with provable convergence to critical points. Overall, this framework shows that a controlled relaxation of convexity enables the design of learnable priors that achieve strong empirical performance while preserving mathematical guarantees.
Image denoising is a fundamental task in image processing, which directly affects the performance of subsequent tasks such as feature extraction and object recognition. Traditional total variation (TV) denoising models are effective in preserving edges but suffer from staircase effects, sensitivity to regularization parameters, and complex numerical solving. Existing Physics-Informed Neural Network (PINN) denoising methods often rely on specific physical constraints, which fail to optimize the image structure, leading to blurred edges and loss of detail. In this paper, we propose a new unsupervised image denoising method based on an improved TV constraint in PINNs. The method takes noisy images as input without requiring clean images as labels, and it achieves superior performance through four key innovations: 1) An improved TV constraint with an adaptive gradient smoothing factor and dynamic regularization parameter to alleviate staircase effects; 2) A generalized physical constraint based on the continuity of image grayscale values, eliminating dependence on specific physical equations; 3) A four-dimensional loss function combining data fidelity, physical consistency, edge preservation, and local variance, with an adaptive weight mechanism to balance the optimization priorities; 4) A two-stage training strategy with an adaptive deep multi-layer perceptron (AD-MLP) to balance convergence speed and model accuracy. Experiments on classic grayscale images with Gaussian noise (variance 0.01–0.05) demonstrate that the proposed method achieves a PSNR improvement of 2.1–5.3 dB, SSIM improvement of 0.06–0.18, and edge preservation index (EPI) improvement of 0.12–0.16, outperforming traditional TV models, pure PINN, Wiener filtering, adaptive TV, and two-stage fusion methods. It significantly mitigates staircase effects and shows superior denoising robustness, structural fidelity, and computational efficiency in complex noise environments.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.