Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Accelerated Full Waveform Inversion by Deep Compressed Learning

Maayan Gelboim, Amir Adler, Mauricio Araya-Polo

TL;DR

The paper tackles the data-volume barrier in Full Waveform Inversion (FWI) by introducing Deep Compressed Learning (DCL) with a binarized sensing layer to learn efficient data acquisition layouts and a representation-learning pipeline (autoencoder) coupled with K-means clustering to select the most informative shot gathers for inversion. The authors propose a two-stage Deep CL–RL framework that first learns candidate compressed sensing patterns and then online selects the best pattern for a given dataset, while simultaneously extracting latent representations for clustering. This work also pioneers the use of representation learning for seismic data in learned latent space to guide data selection for inversion. On a 2D dataset of 5{,}800 layered velocity models, the DCL–RL method consistently outperforms random shot selection, with pronounced gains at low sensing rates (e.g., 10%), enabling faster inversion and paving the way toward accelerated large-scale 3D FWI.

Abstract

We propose and test a method to reduce the dimensionality of Full Waveform Inversion (FWI) inputs as computational cost mitigation approach. Given modern seismic acquisition systems, the data (as input for FWI) required for an industrial-strength case is in the teraflop level of storage, therefore solving complex subsurface cases or exploring multiple scenarios with FWI become prohibitive. The proposed method utilizes a deep neural network with a binarized sensing layer that learns by compressed learning a succinct but consequential seismic acquisition layout from a large corpus of subsurface models. Thus, given a large seismic data set to invert, the trained network selects a smaller subset of the data, then by using representation learning, an autoencoder computes latent representations of the data, followed by K-means clustering of the latent representations to further select the most relevant data for FWI. Effectively, this approach can be seen as a hierarchical selection. The proposed approach consistently outperforms random data sampling, even when utilizing only 10% of the data for 2D FWI, these results pave the way to accelerating FWI in large scale 3D inversion.

Accelerated Full Waveform Inversion by Deep Compressed Learning

TL;DR

The paper tackles the data-volume barrier in Full Waveform Inversion (FWI) by introducing Deep Compressed Learning (DCL) with a binarized sensing layer to learn efficient data acquisition layouts and a representation-learning pipeline (autoencoder) coupled with K-means clustering to select the most informative shot gathers for inversion. The authors propose a two-stage Deep CL–RL framework that first learns candidate compressed sensing patterns and then online selects the best pattern for a given dataset, while simultaneously extracting latent representations for clustering. This work also pioneers the use of representation learning for seismic data in learned latent space to guide data selection for inversion. On a 2D dataset of 5{,}800 layered velocity models, the DCL–RL method consistently outperforms random shot selection, with pronounced gains at low sensing rates (e.g., 10%), enabling faster inversion and paving the way toward accelerated large-scale 3D FWI.

Abstract

We propose and test a method to reduce the dimensionality of Full Waveform Inversion (FWI) inputs as computational cost mitigation approach. Given modern seismic acquisition systems, the data (as input for FWI) required for an industrial-strength case is in the teraflop level of storage, therefore solving complex subsurface cases or exploring multiple scenarios with FWI become prohibitive. The proposed method utilizes a deep neural network with a binarized sensing layer that learns by compressed learning a succinct but consequential seismic acquisition layout from a large corpus of subsurface models. Thus, given a large seismic data set to invert, the trained network selects a smaller subset of the data, then by using representation learning, an autoencoder computes latent representations of the data, followed by K-means clustering of the latent representations to further select the most relevant data for FWI. Effectively, this approach can be seen as a hierarchical selection. The proposed approach consistently outperforms random data sampling, even when utilizing only 10% of the data for 2D FWI, these results pave the way to accelerating FWI in large scale 3D inversion.
Paper Structure (8 sections, 5 equations, 3 figures, 3 tables, 1 algorithm)

This paper contains 8 sections, 5 equations, 3 figures, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Theoretical Background
  3. Deep Compressed Leaning in FWI
  4. Dataset Preparation
  5. Deep Learning Architectures and Training
  6. Performance Evaluation
  7. Conclusions
  8. ACKNOWLEDGMENTS

Figures (3)

  • Figure 1: The proposed DCL-RL solution evaluates sensing patterns, obtained from trained deep compressed learning models. Using the latent representations of the shot gathers, we assign scores to each learned sensing pattern to identify the most informative subset of shot gathers for reconstructing a velocity model.
  • Figure 2: FWI results: per each velocity model (A-H) the ground truth and inversion using all (20) shot gathers are compared vs. inversion with 4 shots (20$\%$ sensing rate), 3 shots (15$\%$) and 2 shots (10$\%$), clearly indicating the advantage of the proposed DCL-RL approach over random sampling.
  • Figure 3: FWI running time on AMD Genoa-X multi-core vs. number of utilized shot gathers.