Research map / Probabilistic and scientific modeling / Molecules and proteins
Physics-Driven ML-Based Modelling for Correcting Inverse Estimation
Ruiyuan Kang, Tingting Mu, Panos Liatsis, Dimitrios Kyritsis
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 |
|---|---|---|---|
| estimate error sum with a single-output MLP instead of per-element errors (ablation) | GEESE · Problem 1, eps=0.05 | average query times: 20.20 -> 23.26 | Hurt |
| train the exploration generator instead of keeping random weights (ablation) | GEESE · Problem 1, eps=0.05 | average query times: 32.64 -> 41.32 | Hurt |
| remove early stopping of base networks (ablation) | GEESE · Problem 1, eps=0.05 | average query times: 20.20 -> 32.80 | Hurt |
| remove focus coefficient (ablation) | GEESE · Problem 1, eps=0.05 | average query times: 20.20 -> 27.19 | Hurt |
| use GEESE instead of GA on Problem 3 | GEESE · Problem 3, eps=0.075 | failure times: 13 -> 0 | Improved |
| use GEESE instead of BOGP on Problem 2 | GEESE · Problem 2, eps=0.075 | failure times: 97 -> 0 | Improved |
| use GEESE instead of PSO on Problem 2 | GEESE · Problem 2, eps=0.075 | average query times: 64.00 -> 51.65 | Improved |
| use GEESE instead of CMAES on Problem 3 | GEESE · Problem 3, eps=0.075 | average query times: 227.42 -> 43.56 | Improved |
| use GEESE instead of BOGP on Problem 1 | GEESE · Problem 1, eps=0.075 | average query times: 3.29 -> 3.18 | No clear effect |
| use SVPEN (RL-based) instead of GEESE | SVPEN · all problems, eps=0.075 | failure times: 0 -> 100 | Hurt |
| lower the feasibility threshold from 0.1 to 0.05 | GEESE · Problem 3 | average query times: 35.58 -> 81.26 | Hurt |
| use GEESE instead of BOGP at eps=0.05 on Problem 1 | GEESE · Problem 1, eps=0.05 | average query times: 9.24 -> 20.20 | Hurt |
Closest papers by meaning
- PIED: Physics-Informed Experimental Design for Inverse Problems (ICLR 2025 · similarity 0.85)
- AFBench: A Large-scale Benchmark for Airfoil Design (NeurIPS 2024 · similarity 0.84)
- 3DID: Direct 3D Inverse Design for Aerodynamics with Physics-Aware Optimization (NeurIPS 2025 · similarity 0.84)
- Model-Agnostic Knowledge Guided Correction for Improved Neural Surrogate Rollout (ICLR 2025 · similarity 0.83)
- GAIA: Geometry-Adaptive Operator Learning for Forward and Inverse Problems (NeurIPS 2026 · similarity 0.83)
- Conformal Inverse Optimization (NeurIPS 2024 · 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.
