Research map / Architectures and efficiency / Attention and transformers
Transformer Meets Boundary Value Inverse Problems
Ruchi Guo, Shuhao Cao, Long Chen
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 convolutions with integral kernel attention (UIT) | U-Net baseline · EIT benchmark tau=0 | Relative L2 error: 0.200 -> 0.163 | Improved |
| replace U-Net convolutions with integral kernel attention (UIT) | U-Net baseline · EIT benchmark tau=0.05 | Relative L2 error: 0.341 -> 0.261 | Improved |
| use L=3 data pairs instead of L=1 (ablation) | UIT · EIT benchmark tau=0 | Relative L2 error: 0.163 -> 0.147 | Improved |
| attention only on coarsest level (U-Net+Coarse Attn) (ablation) | U-Net · EIT benchmark tau=0 | Relative L2 error: 0.200 -> 0.184 at tau=0; 0.341 -> 0.343 at tau=0.05 | Mixed |
| FNO-type global spectral kernels vs attention-based models | FNO2d · EIT benchmark tau=0 | Relative L2 error: 0.318 vs 0.200 for U-Net | Hurt |
| UIT with softmax normalization (UIT+Softmax) vs UIT without softmax (ablation) | UIT · EIT benchmark tau=0 | Relative L2 error: 0.159 vs 0.163 at tau=0; 0.269 vs 0.272 at tau=0.2 | Mixed |
Closest papers by meaning
- Neural Field Thermal Tomography: A Differentiable Physics Framework for Non-Destructive Evaluation (NeurIPS 2026 · similarity 0.84)
- On the Learn-to-Optimize Capabilities of Transformers in In-Context Sparse Recovery (ICLR 2025 · similarity 0.83)
- DMPlug: A Plug-in Method for Solving Inverse Problems with Diffusion Models (NeurIPS 2024 · similarity 0.83)
- Physics-Informed Deep Inverse Operator Networks for Solving PDE Inverse Problems (ICLR 2025 · similarity 0.82)
- Self-diffusion for Solving Inverse Problems (NeurIPS 2025 · similarity 0.82)
- GAIA: Geometry-Adaptive Operator Learning for Forward and Inverse Problems (NeurIPS 2026 · similarity 0.82)
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.
