Research map / Learning theory and algorithms / Optimal transport
Training neural operators to preserve invariant measures of chaotic attractors
Ruoxi Jiang, Peter Y. Lu, Elena Orlova, Rebecca Willett
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 |
|---|---|---|---|
| Add OT loss to RMSE training | FNO · Lorenz-96 r=0.3 | L1 histogram error: 0.215 -> 0.057 | Improved |
| Add contrastive feature loss to RMSE training | FNO · Lorenz-96 r=0.3 | L1 histogram error: 0.215 -> 0.132 | Improved |
| Add contrastive feature loss to RMSE training | FNO · Kuramoto-Sivashinsky r=0.3 | energy spectrum error: 0.290 -> 0.176 | Improved |
| Add OT loss to RMSE training | FNO · Kuramoto-Sivashinsky r=0.3 | L1 histogram error: 0.390 -> 0.172 | Improved |
| Add contrastive loss at high noise | FNO · Lorenz-96 r=0.3 | leading Lyapunov exponent error: 0.440 -> 0.064 | Improved |
| Add contrastive loss at high noise | FNO · Lorenz-96 r=0.3 | fractal dimension error: 3.580 -> 1.894 | Improved |
| Add OT loss at low noise | FNO · Lorenz-96 r=0.1 | leading Lyapunov exponent error: 0.013 -> 0.050 | Hurt |
| Add OT loss on KS data | FNO · Kuramoto-Sivashinsky r=0.3 | leading Lyapunov exponent error: 0.101 -> 0.094 | No clear effect |
| Reduce OT summary statistics from full to partial (ablation) | OT + RMSE · Lorenz-96 r=0.3 | L1 histogram error: 0.057 -> 0.090 | Hurt |
| Reduce OT summary statistics to a non-informative minimum (ablation) | OT + RMSE · Lorenz-96 r=0.3 | L1 histogram error: 0.057 -> 0.221 | Hurt |
| Replace Sinkhorn OT with MMD (ablation) | OT + RMSE · Kuramoto-Sivashinsky r=0.3 | L1 histogram error: 0.172 -> 0.245 | Hurt |
| Replace RMSE with Sobolev norm and dissipative loss | FNO · Lorenz-96 r=0.3 | L1 histogram error: 0.215 -> 0.246 | Hurt |
Closest papers by meaning
- DySLIM: Dynamics Stable Learning by Invariant Measure for Chaotic Systems (ICML 2024 · similarity 0.89)
- Chaos Meets Attention: Transformers for Large-Scale Dynamical Prediction (ICML 2025 · similarity 0.89)
- Learning invariant representations of time-homogeneous stochastic dynamical systems (ICLR 2024 · similarity 0.88)
- When are dynamical systems learned from time series data statistically accurate? (NeurIPS 2024 · similarity 0.88)
- Attractor Memory for Long-Term Time Series Forecasting: A Chaos Perspective (NeurIPS 2024 · similarity 0.87)
- Learning semilinear neural operators: A unified recursive framework for prediction and data assimilation. (ICLR 2024 · 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.
