Research map / Probabilistic and scientific modeling / Molecules and proteins
Learning the Geometry of Wave-Based Imaging
Konik Kothari, Maarten de Hoop, Ivan Dokmanić
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 |
|---|---|---|---|
| Replace U-Net with FIONet | Reverse time continuation · Shapes, Reflectors, MNIST, and CelebA faces; 3000 training samples | normalized cross-correlation: 0.21 -> 0.81; 0.07 -> 0.63; 0.35 -> 0.87; 0.27 -> 0.83 | Improved |
| Replace U-Net with FIONet | Reverse time continuation · Thick lines, 3000 samples | normalized cross-correlation: 0.92 -> 0.97 | Improved |
| Replace U-Net with FIONet | Inverse source problem · Thick lines, Reflectors, Random shapes, and MNIST | normalized cross-correlation: 0.60 -> 0.68; 0.38 -> 0.87; 0.53 -> 0.67; 0.43 -> 0.58 | Improved |
| Replace U-Net with FIONet | Reflector imaging · Reflectors, Shuffled Reflectors, and Random Reflectors | normalized cross-correlation: 0.42 -> 0.74; 0.41 -> 0.72; 0.38 -> 0.55 | Improved |
| Replace U-Net with FIONet | Reflector imaging · Lines | normalized cross-correlation: 0.77 -> 0.75 | Hurt |
| Use FIONet rather than U-Net | Reflector imaging · Lines | smoothed Wasserstein-2: 5.69 -> 6.74 | Hurt |
| Use FIONet rather than U-Net | Reverse time continuation · Out-of-distribution datasets | normalized cross-correlation: Shapes: 0.21 -> 0.81; Reflectors: 0.07 -> 0.63; MNIST: 0.35 -> 0.87; CelebA faces: 0.27 -> 0.83 | Improved |
| Remove the learned multipliers (ablation) | FIONet · All problems | performance | No clear effect |
| Pretrain the routing network before full FIONet training (ablation) | FIONet · Inverse imaging tasks | reconstruction quality | Improved |
Closest papers by meaning
- A Unified Framework for Forward and Inverse Problems in Subsurface Imaging using Latent Space Translations (ICLR 2025 · similarity 0.85)
- Neural Operators with Localized Integral and Differential Kernels (ICML 2024 · similarity 0.85)
- WaveMo: Learning Wavefront Modulations to See Through Scattering (CVPR 2024 · similarity 0.84)
- Exploring Invariance in Images through One-way Wave Equations (ICML 2025 · similarity 0.83)
- Symmetric Basis Convolutions for Learning Lagrangian Fluid Mechanics (ICLR 2024 · similarity 0.83)
- Auto-Linear Phenomenon in Subsurface Imaging (ICML 2024 · similarity 0.83)
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.
