A core research and training program of the GeoSignal Research Institute
This program introduces advanced machine learning techniques specifically tailored for seismic data processing. Emphasis is placed on practical implementation, physical validity, and research-oriented applications for modern industry challenges.
The training bridges classical physical wavefield operators with modern data-driven methodologies, including deep learning and hybrid physics-informed neural networks (PINNs) that respect structural wave equations.
Core thematic pillars covered throughout the training modules:
Advanced deep learning architectures for random/coherent noise attenuation and spatial data reconstruction.
CNNs, Transformers, and Autoencoders customized for 2D/3D seismic waveform interpretation.
Integrating wave equation constraints into deep neural networks to guarantee physical consistency.
Real-world deployment scenarios across offshore exploration, DAS monitoring, and microseismic tracking.
Teaching resources are organized to progressively build foundational knowledge up to deploying reproducible workflows on complex 2D/3D seismic datasets.
Designed for graduate students, geophysicists, PhD researchers, and industry professionals looking to integrate state-of-the-art machine learning algorithms into modern seismic acquisition, processing, and interpretation pipelines.