Brook Preloader

Machine Learning for low-frequency seismic measurement

Led by Dr Velimir Ilić (MISANU, Serbia) and Fabio Bonsignorio (Heron Robots, Italy)

The performance of Earth-based GW detectors is largely influenced by the ability of combating the low-frequency ambient seismic noise and other seismic influences. These tasks require multidisciplinary research in the fields of seismic sensing, signal processing, robotics, machine learning and mathematical modeling.

The working group 2, deals with acquisition, processing and interpretation of seismic data, with the goal of combating the seismic influences at GW detector site, using the multidisciplinary research, with a focus on advanced techniques available from state of the art machine learning algorithms.

WG2 is actively working on the following tasks

• Nonstationary signal analysis: Extraction of Useful Information from Noisy Signals; GW and seismic signal denoising; Time-Frequency Representations of seismic data.
• Deep learning for seismic time series: Rapid prediction of earthquake ground shaking intensity for VIRGO using ML; Intra-domain and cross-domain transfer learning for time series; Learning to Invert Pseudo-Spectral Data for Seismic Waveform Inversion.
• Newtonian noise detection and seismic sensor arrays: Analysis of seismic data from the Virgo network; Seismic networks at the ET candidate site in Sardinia; Seismic networks for machine learning, Advanced equipment (mobile seismic sensors, infrasound microphone…).
• Probabilistic robotics for seismology and GW detection: Gaussian and non-Gaussian processes for the characterization of Newtonian, seismic and other kinds of noises; Advanced information theoretic methods for multi-sensory fusion to drive the information gathering and motion planning (information theoretic measures, probabilistic graphical models, causal networks, dynamic programming…); Simultaneous localization and geographical fields mapping by means of network or swarms of autonomous mobile robots.

Check for our recent EGU 2020 talk: Ambient seismic noise suppression in COST action G2Net For more information (working plans, meeting notes, state of the art references, etc.), and to access the WG2 online meetings, feel free to join us and subscribe to our mailing list: wg2-g2net at or contact the coordinators of this WG whose emails can be found here.

WG2 members:
Jonatan Lerga: jlerga at riteh dot hr
Christopher Zerafa: christopher dot zerafa dot 08 at um dot edu dot mt
Ivan Štajduhar: istajduh at riteh dot hr
Dario Jozinović: djozinovi at gmail dot com
Tomek Builk: tb at astrouw dot edu dot pl
Fabio Bonsignorio: fabio dot bonsignorio at gmail dot com
Velimir Ilić: velimir dot ilic at gmail dot com


The inclusion of further partners from currently participating COST Countries, or other countries within or outside the COST Network, is welcome and strongly encouraged during the entire duration of the Action.