Research map / Architectures and efficiency / RNNs and normalization
Online Deep Equilibrium Learning for Regularization by Denoising
Jiaming Liu, Xiaojian Xu, Weijie Gan, Shirin Shoushtari, Ulugbek Kamilov
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use stochastic minibatches of measurements in the DEQ forward and backward passes instead of full-batch measurements | RED (DEQ) · IDT, 15 dB input SNR | SNR (dB): 40.54 -> 40.57 | No clear effect |
| Use stochastic minibatches of measurements in the DEQ forward and backward passes instead of full-batch measurements | RED (DEQ) · Sparse-view CT, 90 views | SNR (dB): 34.61 -> 34.40 | Hurt |
| Use stochastic minibatches of measurements in the DEQ forward and backward passes instead of full-batch measurements | RED (DEQ) · MRI Set1, 10% sampling | SNR (dB): 24.10 -> 24.08 | No clear effect |
| Use stochastic minibatches of measurements in the DEQ forward and backward passes instead of full-batch measurements | RED (DEQ) · IDT, 20 dB input SNR | GPU test time: 6.362s -> 2.051s | Improved |
| Use stochastic minibatches of measurements in the DEQ forward and backward passes instead of full-batch measurements | RED (DEQ) · Sparse-view CT, 180 views | GPU test time: 5.466s -> 3.113s | Improved |
| Use stochastic minibatches of measurements in the DEQ forward and backward passes instead of full-batch measurements | RED (DEQ) · MRI, 320×320 image, 96 coils | GPU test time: 4.577s -> 3.005s | Improved |
| Use stochastic minibatches of measurements in the DEQ forward and backward passes instead of full-batch measurements | RED (DEQ) · IDT, 20 dB input SNR | measurement memory per iteration: 3.56 GB -> 0.71 GB | Improved |
| Increase the ODER measurement minibatch from 100 to 250 (ablation) | ODER · IDT, 15 dB input SNR | SNR (dB): 40.28 -> 40.57 | Improved |
Closest papers by meaning
- Provably Robust Score-Based Diffusion Posterior Sampling for Plug-and-Play Image Reconstruction (NeurIPS 2024 · similarity 0.89)
- i-DEQ: A stable inertial Deep Equilibrium model for image restoration (NeurIPS 2026 · similarity 0.89)
- Plug-and-Play image restoration with Stochastic deNOising REgularization (ICML 2024 · similarity 0.88)
- A Variational Perspective on Solving Inverse Problems with Diffusion Models (ICLR 2024 · similarity 0.87)
- Prior Mismatch and Adaptation in PnP-ADMM with a Nonconvex Convergence Analysis (ICML 2024 · similarity 0.87)
- Learning Cocoercive Conservative Denoisers via Helmholtz Decomposition for Poisson Imaging Inverse Problems (NeurIPS 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
