Research map / Learning theory and algorithms / Matrix and tensor methods
Robust compressed sensing using 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 |
|---|---|---|---|
| Replace squared-loss ERM with the MOM tournament objective | Generative-model compressed sensing · Heavy-tailed samples with arbitrary corruptions | reconstruction error | Helped |
| Replace squared-loss ERM with the MOM tournament objective | Generative-model compressed sensing · Heavy-tailed noise without arbitrary outliers | reconstruction error | Helped |
| Use MOM minimization instead of the MOM tournament objective (ablation) | MOM-based generative-model reconstruction · CelebA-HQ, heavy-tailed measurements and arbitrary corruptions | reconstruction error | No clear effect |
| Replace MOM reconstruction with ERM | Generative-model compressed sensing · Sub-Gaussian measurements without corruptions | MSE | Helped |
| Reduce the number of batches in MOM (ablation) | MOM reconstruction · Sub-Gaussian measurements without corruptions | MSE | Helped |
| Replace the MOM tournament with ERM | Generative-model compressed sensing · Heavy-tailed measurement matrix without corruptions | reconstruction error | No clear effect |
| Replace squared loss with L1 loss minimization | Generative-model reconstruction · CelebA-HQ, heavy-tailed measurements and arbitrary corruptions | reconstruction error | Hurt |
| Use trimmed loss minimization instead of MOM | Generative-model reconstruction · CelebA-HQ, heavy-tailed measurements and arbitrary corruptions | reconstruction error | Hurt |
Closest papers by meaning
- Compressed Sensing using Generative Models (ICML 2017 · similarity 0.91)
- Modeling Sparse Deviations for Compressed Sensing using Generative Models (ICML 2018 · similarity 0.90)
- Robust One-Bit Recovery via ReLU Generative Networks: Near-Optimal Statistical Rate and Global Landscape Analysis (ICML 2020 · similarity 0.90)
- Instance-Optimal Compressed Sensing via Posterior Sampling (ICML 2021 · similarity 0.90)
- Sparse Bayesian Generative Modeling for Compressive Sensing (NeurIPS 2024 · similarity 0.87)
- Recovery Guarantees for Posterior Sampling of One-Bit Compressed Sensing (NeurIPS 2026 · 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.