Notice: The reproducibility variables underlying each score are classified using an automated LLM-based pipeline, validated against a manually labeled dataset. LLM-based classification introduces uncertainty and potential bias; scores should be interpreted as estimates. Full accuracy metrics and methodology are described in Coakley et alK. L. Coakley, T. Snelleman, H. Hoos, and O. E. Gundersen, "The embrace of open science: An analysis of a decade of AI research and 56 800 conference papers," Under Review, 2026..
PhySense: Sensor Placement Optimization for Accurate Physics Sensing
Authors: Yuezhou Ma, Haixu Wu, Hang Zhou, Huikun Weng, Jianmin Wang, Mingsheng Long
NeurIPS 2025 | Venue PDF | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experiments across three challenging benchmarks, especially a 3D geometry dataset, indicate Phy Sense achieves state-of-the-art physics sensing accuracy and discovers informative sensor placements previously unconsidered. |
| Researcher Affiliation | Academia | Yuezhou Ma, Haixu Wu, Hang Zhou, Huikun Weng, Jianmin Wang, Mingsheng Long School of Software, BNRist, Tsinghua University, China EMAIL, EMAIL, EMAIL |
| Pseudocode | No | The paper describes the methods in natural language and mathematical equations but does not include explicit pseudocode or algorithm blocks. |
| Open Source Code | Yes | Code is available at this repository: https://github.com/thuml/Phy Sense. |
| Open Datasets | Yes | We evaluate our method across three challenging benchmarks, including turbulent flow simulations, reanalysis of global sea temperature, and industrial simulations of aerodynamic surface pressure over a 3D car on an irregular geometry... The GLORYS12 reanalysis dataset [21]... from Shape Net [6] |
| Dataset Splits | Yes | Our training set comprises 9,843 daily samples from 1993 to 2019, with data from 2020 2021 reserved for testing... We simulate 100 cases in total, splitting them into 75 for training and 25 for testing. |
| Hardware Specification | Yes | All the experiments are conducted based on Py Torch 2.1.0 [34] and on a A100 GPU server with 144 CPU cores. |
| Software Dependencies | Yes | All the experiments are conducted based on Py Torch 2.1.0 [34] |
| Experiment Setup | Yes | Table 8: Common setups for all models in three benchmarks. Other setups follow their original setup. Benchmarks Turbulent Flow Sea Temperature Car Aerodynamics Epochs 800 300 300 Batch Size 60 40 1 Learning Rate 10 3 2 10 4 10 3 Optimizer ADAM... Table 11: Learning rate of sensor placement optimization on three benchmarks. Benchmarks Turbulent Flow Sea Temperature Car Aerodynamics Learning rate 0.25 1 0.0025 |