An advanced generalized autoencoder for dimensionality reduction and feature extraction
AutoencoderZ is an advanced Autoencoder model designed for dimensionality reduction of various data types, such as seismometer and strainmeter data. It features an encoder-decoder architecture that efficiently compresses and reconstructs input data while preserving important features. The extracted features can then be used for various applications, such as unsupervised clustering, anomaly detection, and pattern recognition.
When using this model, please cite the following:
Zali, Z., Mousavi, S.M., Ohrnberger, M. et al. Tremor clustering reveals pre-eruptive signals and evolution of the 2021 Geldingadalir eruption of the Fagradalsfjall Fires, Iceland. Commun Earth Environ 5, 1 (2024). https://doi.org/10.1038/s43247-023-01166-w
Zali, Z., Martínez-Garzón,P., Kwiatek, G., Núñez-Jara, S., Beroza, G., Cotton, F., Bohnhoff, M. Low-Frequency Tremor-Like Episodes Before the 2023 MW 7.8 Türkiye Earthquake Linked to Cement Quarrying. (under review with Scientific Reports)
Python 3, TensorFlow 2, Keras, NumPy
Feel free to open issues or submit pull requests to enhance the functionality or improve the code.
Developer: Zahra Zali, [email protected]