HSDL A novel and practical method is proposed to refine an automatic earthquake catalog through the integration of hybrid shallow and deep learning techniques. The approach leverages the complementary strengths of both paradigms: shallow learning models, such as gradient boosting and support vector machines, offer interpretability, computational efficiency, and robustness for structured seismic attributes, while deep neural networks capture complex temporal–spatial patterns and waveform features that are often overlooked by traditional algorithms. By combining these two layers of intelligence, the method enables the systematic identification and correction of misclassified or low-quality events in large-scale automatic catalogs. The hybrid framework not only enhances the precision of event detection and phase association but also reduces false positives, ensuring a more reliable and consistent dataset for subsequent seismological analyses. This fusion of shallow and deep learning represents a significant step toward automated, data-driven catalog refinement, bridging the gap between algorithmic efficiency and geophysical interpretability.
Siervo et al. (2025). HSDL: A novel and practical method to refine an automatic earthquake catalog using hybrid shallow and deep learning, TBD.
BibTeX:
@article{huang2016dmssa,
title={HSDL: A novel and practical method to refine an automatic earthquake catalog using hybrid shallow and deep learning},
author={Siervo et al.},
journal={TBD},
volume={TBD},
number={TBD},
issue={TBD},
pages={TBD},
year={2026}
}
HSDL developing team, 2024-present
MIT License
Using the latest version
git clone https://github.com/chenyk1990/HSDL
cd HSDL
pip install -v -e .
The "demo" directory contains all runable scripts to demonstrate different applications of HSDL.
The gallery figures of the HSDL package can be found at https://github.com/chenyk1990/gallery/tree/main/HSDL Each figure in the gallery directory corresponds to a DEMO script in the "demo" directory with the exactly the same file name.
- scipy
- numpy
- matplotlib
The development team welcomes voluntary contributions from any open-source enthusiast.
If you want to make contribution to this project, feel free to contact the development team.
Regarding any questions, bugs, developments, or collaborations, please contact
Yangkang Chen
chenyk2016@gmail.com
The gallery figures of the HSDL package can be found at https://github.com/chenyk1990/gallery/tree/main/hsdl
Generated by notebooks/DEMO_test1.ipynb
Generated by notebooks/DEMO_test2.ipynb
Generated by notebooks/DEMO_test1.ipynb
Generated by notebooks/DEMO_test2.ipynb
Generated by notebooks/DEMO_test1.ipynb
Generated by notebooks/DEMO_test2.ipynb
Generated by notebooks/DEMO_test1.ipynb
Generated by notebooks/DEMO_test2.ipynb
Generated by notebooks/DEMO_test1.ipynb
Generated by notebooks/DEMO_test2.ipynb
Generated by notebooks/DEMO_test1.ipynb
Generated by notebooks/DEMO_test2.ipynb
Generated by notebooks/DEMO_test1.ipynb
Generated by notebooks/DEMO_test2.ipynb
