Research map / Image generation / Image restoration
Intermediate Layer Optimization for Inverse Problems using Deep Generative Models
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Add stochastic noise addition (SNA) (ablation) | ILO · Noisy image (σ=30) | PSRN (dB): 28.34 -> 32.92 | Helped |
| Add stochastic noise addition (SNA) (ablation) | CSGM · Noisy image (σ=30) | PSRN (dB): 19.89 -> 21.38 | Helped |
| Use ILO instead of the best CSGM-based baseline | StyleGAN-2 inverse reconstruction · Denoising (σ=25) | reconstruction error: 1.8× better | Helped |
| Use ILO instead of PULSE (CSGM + MSE) | StyleGAN-2 inverse reconstruction · Celeba-HQ super-resolution (1024×1024 to 64×64) | reconstruction error: 1.65× better | Helped |
| Use ILO instead of PULSE (CSGM + MSE) | StyleGAN-2 inverse reconstruction · Celeba-HQ super-resolution (32×32 images) | reconstruction error: 1.4× better | Helped |
| Use ILO instead of CSGM | StyleGAN-2 inverse reconstruction · Inpainting (5%–100% observed pixels) | reconstruction error: 2× better | Helped |
| Use ILO instead of CSGM | StyleGAN-2 inverse reconstruction · Compressed sensing with partial circulant matrices (5% of dimension n) | reconstruction error: 2× better | Helped |
Closest papers by meaning
- Invertible generative models for inverse problems: mitigating representation error and dataset bias (ICML 2020 · similarity 0.89)
- Score-Guided Intermediate Level Optimization: Fast Langevin Mixing for Inverse Problems (ICML 2022 · similarity 0.89)
- Differentiable Gaussianization Layers for Inverse Problems Regularized by Deep Generative Models (ICLR 2023 · similarity 0.89)
- Integrating Intermediate Layer Optimization and Projected Gradient Descent for Solving Inverse Problems with Diffusion Models (ICML 2025 · similarity 0.88)
- Solving Inverse Problems with FLAIR (NeurIPS 2025 · similarity 0.87)
- pcaGAN: Improving Posterior-Sampling cGANs via Principal Component Regularization (NeurIPS 2024 · 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.