Research map / Label-efficient and robust learning / Self-supervised and contrastive
Self-Supervised Learning with Lie Symmetries for Partial Differential Equations
Grégoire Mialon, Quentin Garrido, Hannah Lawrence, Danyal Rehman, Yann LeCun, Bobak Kiani
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 |
|---|---|---|---|
| pretrain with SSL and linear head instead of supervised training | ResNet18 · KdV initial condition regression | NMSE: 0.102 -> 0.033 | Improved |
| pretrain with SSL instead of supervised | ResNet18 · KS initial condition regression | NMSE: 0.117 -> 0.042 | Improved |
| pretrain with SSL instead of supervised | ResNet18 · Burgers viscosity | relative error %: 1.18 -> 0.97 | Improved |
| pretrain with SSL instead of supervised | ResNet18 · Navier-Stokes buoyancy | MSE: 0.0078 -> 0.0038 | Improved |
| condition time-stepping network on SSL representation | CNN · KdV | NMSE: 0.508 -> 0.330 | Improved |
| condition time-stepping network on SSL representation | CNN · KS | NMSE: 0.549 -> 0.381 | Improved |
| condition time-stepping network on SSL representation | CNN · Burgers | NMSE: 0.110 -> 0.108 | No clear effect |
| add SSL conditioning to UNet with AdaGN | UNet_mod64 · Navier-Stokes | MSE x1e3: 2.37 -> 2.35 | No clear effect |
| add SSL conditioning to UNet with Addition | UNet_mod64 · Navier-Stokes | MSE x1e3: 2.60 -> 2.47 | Improved |
| add SSL conditioning to UF1Net | UF1Net · Navier-Stokes | MSE x1e3: 3.31 -> 2.37 | Improved |
| add SSL conditioning to FNO | FNO · Navier-Stokes | MSE x1e3: 13.4 -> 13.0 | No clear effect |
| use true buoyancy instead of SSL representation (ablation) | UNet_mod64 AdaGN · Navier-Stokes | MSE x1e3: 2.35 -> 2.01 | Improved |
Closest papers by meaning
- Data-Efficient Operator Learning via Unsupervised Pretraining and In-Context Learning (NeurIPS 2024 · similarity 0.88)
- Graph Neural PDE Solvers with Conservation and Similarity-Equivariance (ICML 2024 · similarity 0.87)
- Mechanistic PDE Networks for Discovery of Governing Equations (ICML 2025 · similarity 0.87)
- Neural Spectral Methods: Self-supervised learning in the spectral domain (ICLR 2024 · similarity 0.87)
- Solving Differential Equations with Constrained Learning (ICLR 2025 · similarity 0.87)
- Active Learning for Neural PDE Solvers (ICLR 2025 · similarity 0.86)
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.
