📝 Publications
JGR M&C
Automatic differentiation‐based full waveform inversion with flexible workflows
Liu, F., Li, H., Zou, G., & Li, J. (2025)
🔗Code Repository: github.com/liufeng2317/ADFWI
- ADFWI: A PyTorch-based full waveform inversion framework leveraging automatic differentiation.
- Multi-Physics: Supports acoustic, elastic, and anisotropic wave propagation.
- Multi-Workflow: Integrates diverse misfit functions, regularization methods, and optimizers for flexible inversion design.
- Multi-Platform: Runs on both CPU and GPU with parallel computing support.
Arxiv
Deep Reparameterization for Full Waveform Inversion: Architecture Benchmarking, Robust Inversion, and Multiphysics Extension (Version 2)
Liu, F., Li, Y., Su, R., Huang, J., & Bai, L. (2025)
🔗Code Repository: [github.com/liufeng2317/ADFWI]
- DRFWI: A deep reparameterization framework for full waveform inversion (FWI).
- Architecture Benchmarking: Evaluates CNNs, UNets, MLPs, and other architectures for model reparameterization.
- Robustness: Assesses inversion performance under varying noise levels and sparse acquisition.
- Multi-Physics Extension: Introduces a backbone-branch structure for multiparameter inversion, reducing cross-parameter interference.
JGR M&C
DispFormer: A Pretrained Transformer Incorporating Physical Constraints for Dispersion Curve Inversion
Liu, F., Deng, B., Su, R., Bai, L., & Ouyang, W. (2025)
🔗Code Repository: github.com/liufeng2317/DispFormer
- DispFormer: A pretrained transformer for flexible dispersion curve inversion, from global synthesis to regional applications.
- Zero-shot: Directly inverts dispersion curves of arbitrary lengths without target-specific training.
- Few-shot: Outperforms state-of-the-art global search methods with minimal training data.
- Multi-Physics Extension: Easily adaptable to tasks like 1D electrical inversion, receiver function inversion, and multi-modal surface wave inversion.
GJI
Multimodal surface wave inversion with automatic differentiation
Liu, F., Li, J., Fu, L., & Lu, L. (2024)
🔗Code Repository: github.com/liufeng2317/ADsurf
- ADsurf: A PyTorch-based framework for multimodal surface wave inversion using automatic differentiation.
- High-Performance: Optimized using GPU with parallel computing support (10x faster than CPU).
- Uncertainty Quantification: Provides uncertainty estimates for inversion results.
Arxiv
OpenSWI: A Massive-Scale Benchmark Dataset for Surface Wave Dispersion Curve Inversion
Liu, F., Zhao, S., Gu, X., Ling, F., Zhuang, P., Li, Y., Su, R., Fang, L., Zhou, L., Huang, J., & Bai, L. (2025)
🔗Code Repository: github.com/liufeng2317/OpenSWI
- Massive-Scale: 22M shallow and 1.26M deep synthetic samples, plus real-world observations.
- Comprehensive Coverage: From near-surface geology to deep Earth structures.
- Benchmark Standard: Unified dataset for developing and evaluating AI-based inversion methods.
- Open & Reproducible: Generated via the SWIDP pipeline and fully released with code and examples.
🎙 Geophysical Inversion
- Online on
arXiv: Liu, F., Zhao, S., Gu, X., Ling, F., Zhuang, P., Li, Y., Su, R., Fang, L., Zhou, L., Huang, J., & Bai, L. (2025). OpenSWI: A Massive-Scale Benchmark Dataset for Surface Wave Dispersion Curve Inversion - Online on
arXiv: Liu, F., Li, Y., Su, R., Huang, J., & Bai, L. (2025). Deep reparameterization for full waveform inversion: Architecture benchmarking, robust inversion, and multiphysics extension - Accept by
JGR: Machine Learning and Computation: Liu, F., Deng, B., Su, R., Bai, L., & Ouyang, W. (2025). DispFormer: A Pretrained Transformer Incorporating Physical Constraints for Dispersion Curve Inversion - Published in
JGR: Machine Learning and Computation: Liu, F., Li, H., Zou, G., & Li, J. (2025). Automatic differentiation‐based full waveform inversion with flexible workflows - Published in
Geophysical Journal International: Liu, F., Li, J., Fu, L., & Lu, L. (2024). Multimodal surface wave inversion with automatic differentiation
🎙 Seismic Monitoring
- Submitted to
NeurIPS 2025: Wang, F., Chen, M., He, X., Zhang, Y.-F., Liu, F., et al. (2025). OmniEarth-bench: Towards holistic evaluation of Earth’s six spheres and cross-spheres interactions with multimodal observational earth data- Contribution: Manuscript revision and partial contribution to data collection and curation.
- Submitted to
GRL: Wang, X., Liu, F., Su, R., Wang, Z., Bai, L., & Ouyang, W. (2025). SeisMoLLM: Advancing Seismic Monitoring via Cross-modal Transfer with Pre-trained Large Language Model- Contribution: Responsible for manuscript revision and overall coordination.
- Online on
arXiv: Zhang, T., Liu, F., Yuan, Y., Su, R., Ouyang, W., & Bai, L. (2024). Fast Information Streaming Handler (FisH): A Unified Seismic Neural Network for Single Station Real-Time Earthquake Early Warning- Contribution: Designed and implemented the main framework and code.
- Published in
Bulletin of the Seismological Society of America: Han, X., Li, Z., Liu, F., Li, J., & Yao, H. (2025). Real-time local shear-wave splitting measurement: Application to the vicinity of the baihetan hydropower plant- Contribution: Provided the real-time monitoring code.
- Published in
Earthquake Research Advances: Li, J., Yao, H., Wang, B., Yang, Y., Hu, X., Zhang, L., Ye, B., Yang, J., Li, X., Liu, F., Chen, G., Guo, C., & Yang, W. (2022). A real-time AI-assisted seismic monitoring system based on new nodal stations with 4G telemetry and its application in the Yangbi MS 6.4 aftershock monitoring in southwest China- Contribution: Designed and implemented the main framework and code.
🎙 Remote Sensing
- Submitted to
NeurIPS 2025: Zhao, S., Liu, F., Zhang, X., Chen, H., Xiao, P., & Bai, L. (2025). Temporal-Spectral-Spatial Unified Remote Sensing Dense Prediction- Contribution: Methodology, Manuscript writting and revision.
- Submitted to
AAAI 2025: Zhao, S., Liu, F., Zhang, X., Chen, H., Han, T., Gong, J., Tao, R., Xiao, P., Bai, L., & Ouyang, W. (2025). Transforming weather data from pixel to latent space (Version 2)- Contribution: Methodology, Manuscript writting and revision.