Research map / Probabilistic and scientific modeling / Differential equations and PDEs
TINNs: Time-Induced Neural Networks for Solving Time-Dependent PDEs
Chen-Yang Dai, Che-Chia Chang, Te-Sheng Lin, Ming-Chih Lai, Chieh-Hsin Lai
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 time-as-input with time-induced parameters (TINN vs subMLP, same backbone, LM) (ablation) | space-time PINN backbone · Burgers | relative L2 error: 7.11E-05 -> 6.89E-07 | Improved |
| replace time-as-input with time-induced parameters (TINN vs subMLP, same backbone, LM) (ablation) | space-time PINN backbone · Allen-Cahn | relative L2 error: 1.76E-03 -> 3.85E-06 | Improved |
| use compact layer-wise time embedding instead of linear trajectory (ablation) | TINN theta(t) · Burgers | relative L2 error: 2.65E-06 -> 5.67E-07 | Improved |
| use compact layer-wise time embedding instead of one-neuron trajectory (ablation) | TINN theta(t) · Allen-Cahn | relative L2 error: 1.47E-05 -> 2.73E-06 | Improved |
| linear vs one-neuron trajectory for theta(t) (ablation) | TINN theta(t) · Burgers | relative L2 error: 2.65E-06 vs 2.93E-06 | No clear effect |
| TINN time-induced parameterization instead of PirateNet, Adam training (ablation) | TINN · Allen-Cahn | relative L2 error: 2.32E-03 -> 8.29E-05 | Improved |
| TINN trained with Adam instead of LM versus PirateNet (ablation) | TINN · Burgers | relative L2 error: 2.43E-05 -> 2.15E-05 | No clear effect |
| layer-wise temporal embedding instead of naive HyperPINN-style hypernetwork (ablation) | TINN · Klein-Gordon | relative L2 error: 2.02E-01 -> 4.78E-06 | Improved |
| layer-wise temporal embedding instead of naive HyperPINN-style hypernetwork (ablation) | TINN · Allen-Cahn | relative L2 error: 3.49E-06 vs 3.85E-06 | No clear effect |
| time-induced parameterization instead of vanilla PINN | TINN · Allen-Cahn | relative L2 error: 4.65E-01 -> 3.85E-06 | Improved |
| time-induced parameterization versus PirateNet SOAP | TINN · Korteweg-De Vries | relative L2 error: 4.26E-04 -> 1.53E-04 | Improved |
| time-induced parameterization versus CoPINN* on Klein-Gordon | TINN · Klein-Gordon | relative L2 error: 6.61E-06 -> 4.78E-06 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CoPINN* — CoPINN: Cognitive Physics-Informed Neural Networks (ICML 2025) · better on 5 of 5 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- SineNet: Learning Temporal Dynamics in Time-Dependent Partial Differential Equations (ICLR 2024 · similarity 0.88)
- Sub-Sequential Physics-Informed Learning with State Space Model (ICML 2025 · similarity 0.88)
- RandNet-Parareal: a time-parallel PDE solver using Random Neural Networks (NeurIPS 2024 · similarity 0.88)
- PINNsFormer: A Transformer-Based Framework For Physics-Informed Neural Networks (ICLR 2024 · similarity 0.88)
- Neuro-Spectral Architectures for Causal Physics-Informed Networks (NeurIPS 2025 · similarity 0.87)
- Efficient Error Certification for Physics-Informed Neural Networks (ICML 2024 · 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.
