I am a PhD Candidate at King Abdullah University of Science and Technology (KAUST) in Seismic Modeling and Inversion group run by Prof. Daniel Peter. I am broadly interested in numerical methods in geophysics, which includes but not limited to Machine Learning, Seismic Imaging, HPC and Natural Stress State Reconstruction

My PhD research is focused on developing ML-based methods to improve seismic inversion.


Recent research

  • Style transfer for generation of realistically textured subsurface models

    We apply an iterative style transfer approach from image processing to produce realistically textured subsurface models based on synthetic prior models. These realistically textured models to be used in training datasets for machine learning applications in geophysics. Presented in Sep 2019, on SEG Annual meeting in San Antonio, USA

  • Transfer learning for low frequency extrapolation from shot gathers for FWI applications

    Here, we propose and utilize transfer learning to reduce the computational efforts for optimal architecture search and initial network training. We re-train the light-weight MobileNet convolutional network to infer low-frequency data from a frequency-domain representation of individual shot-gathers, which leads to an efficient, yet accurate inference of low fre- quencies according to wavenumber theory. Presented in Jun 2019, on EAGE Annual meeting in London, UK

  • Low-Frequency Data Extrapolation Using a Feed-Forward ANN

    We explored the feasibility of frequency-bandwidth extrapolation using an Artificial Neural Network (ANN) approach. The ANN is trained to be a non-linear operator that maps high-frequency data for a single source and multiple receivers to low-frequency data.

  • Feasibility of moment tensor inversion from a single borehole data using Artificial Neural Networks

    We discuss the feasibility of using artificial neural networks for moment tensor inversion of three-component microseismic data from a single vertical well. For this purpose, we solve a nonlinear regression problem using a feed-forward artificial neural network (ANN).

  • Variance-based automated salt flooding for FWI

    We have developed a variance-based method for reconstruction of velocity models to resolve the imaging and inversion issues caused by salt bodies. Our main idea lies in retrieving useful information from independent updates corresponding to FWI at different frequencies.


News

  • 18/09/19: Vladimir Kazei presented our work on style transfer for generation of realistic subsurface models at SEG Annual Meeting 2019 in San Antonio, USA. Code available!
  • 06/08/19: Started internship as Machine Learnin Engineer Intern at CGG, Crawley, UK.
  • 13/06/19: Awarded SEG ExxonMobil Upstream Research Company scolarship for the year 2019-2020!
  • 11/03/19: Abstract "Transfer learning for low frequency extrapolation from shot gathers for FWI applications" got accepted for an oral presentation on June 6th at EAGE 2019 in London, UK.
  • 29/01/19: Published an easy to understand CUDA code for acoustic wave propagation in a 2D domain. This uses shared memory tiles on device to partially eliminate redundant memory reads
  • 19/10/18: Vladimir Kazei presented our work on frequency frequency-bandwidth extrapolation at SEG 2018 in Anaheim, USA
  • 17/09/18: Released v.1.0 WaveProp in MATLAB. Single-file scripts for 2D and 3D, acoustic and elastic FDTD wave propagation. Intended to be a set of simple codes for beginners
  • 07/09/18: Our paper Variance-based model interpolation for improved full-waveform inversion in the presence of salt bodies got publised in GEOPHYSICS