Neural-networks for geophysicists and their application to seismic data interpretation

@article{Peters2019NeuralnetworksFG,
  title={Neural-networks for geophysicists and their application to seismic data interpretation},
  author={Bas Peters and Eldad Haber and Justin Granek},
  journal={ArXiv},
  year={2019},
  volume={abs/1903.11215}
}
There has been a surge of interest in neural networks for the interpretation of seismic images over the last few years. Network-based learning methods can provide fast and accurate automatic interpretation, provided that there are many training labels. We provide an introduction to the field for geophysicists who are familiar with the framework of forward modeling and inversion. We explain the similarities and differences between deep networks and other geophysical inverse problems and show… 

Figures from this paper

3D seismic interpretation with deep learning: A brief introduction

Understanding the internal structure of our planet is a fundamental goal of the earth sciences. As direct observations are restricted to surface outcrops and borehole cores, we rely on geophysical

Deep learning of geological structures in seismic reflection data

Understanding the internal structure of our planet is a fundamental goal of the Earth Sciences. As direct observations are restricted to surface outcrops and borehole cores, we rely on geophysical

Does shallow geological knowledge help neural-networks to predict deep units?

It is shown that knowledge of shallow geological units helps to predict deeper units when there are only a few labels for training using a dataset from the Sea of Ireland, and U-net based multi-resolution networks can be described using matrix-vector product notation in a similar fashion as standard geophysical inverse problems.

Extracting horizon surfaces from 3D seismic data using deep learning

This work has formulated horizon picking as a multiclass segmentation problem and solved it by supervised training of a 3D convolutional neural network by designing an efficient architecture to analyze the data over multiple scales while keeping memory and computational needs to a practical level.

Deep learning for probabilistic salt segmentation using Bayesian inference machines

  • T. KonukJ. Shragge
  • Geology, Computer Science
    First International Meeting for Applied Geoscience & Energy Expanded Abstracts
  • 2021
A hybrid fully convolutional architecture that combines deterministic and probabilistic layers, which provides an efficient DL methodology for obtaining true salt probabilities and model uncertainties is proposed.

Fully Reversible Neural Networks for Large-Scale 3D Seismic Horizon Tracking

A fully reversible network for horizon tracking that has a memory requirement that is independent of network depth, and which uses the saved memory to increase the input size of the data by order of magnitude such that the network can better learn from large structures in the data.

Automatic Channel Detection Using DNN on 2D Seismic Data

An innovative automatic channel detection algorithm based on machine learning techniques that can identify channels in seismic data/images fully automatically and tremendously increases the efficiency and accuracy of the interpretation process.

Mineral prospectivity mapping using a VNet convolutional neural network

This work tests an artificial intelligence architecture called VNet that uses deep learning and convolutional neural networks for mineral prospectivity on an orogenic gold greenstone belt setting in the Canadian Arctic where the algorithm uses gold values from sparse drill holes for training purposes to predict gold mineralization elsewhere in the region.

Advances in Geo-Time Series Modelling

Recent advances in geo-time series modelling are briefly presented. These progressive developments and imminent applications in the data-driven research have come across three main categories of

Fully reversible neural networks for large-scale surface and sub-surface characterization via remote sensing

This work shows how the cross-entropy loss function requires small modifications to work in conjunction with a fully reversible network and learn from sparsely sampled labels without ever seeing fully labeled ground truth from hyperspectral time-lapse data.

References

SHOWING 1-10 OF 27 REFERENCES

Application of deep learning for seismic horizon interpretation

The interpretation of key horizons on seismic data is an essential but time-consuming part of the subsurface workflow. This is compounded when surfaces need to be re-interpreted on variations of the

Multi-resolution neural networks for tracking seismic horizons from few training images

A projected loss-function for training convolutional networks with a multi-resolution structure, including variants of the U-net is proposed and labels are proposed as the convolution of a Gaussian kernel and the known horizon locations that indicate uncertainty in the labels.

Automatic classification of geologic units in seismic images using partially interpreted examples

The combination of a partial-loss function, a multi-resolution network that explicitly takes small and large-scale geological features into account, and new labeling strategies make neural networks a more practical tool for automatic seismic interpretation.

Convolutional neural networks for automated seismic interpretation

The intuition behind convolutional neural networks, a method revolutionizing the field of image analysis and pushing the state of the art, is looked into and considerations that must be made in order to make the method reliable are discussed.

A deep convolutional encoder-decoder neural network in assisting seismic horizon tracking

Seismic horizons are geologically significant surfaces that can be used for building geology structure and stratigraphy models. However, horizon tracking in 3D seismic data is a time-consuming and

Seismic horizon picking using an artificial neural network

It is shown that this method makes better use of the general properties of horizons, is more robust than conventional pattern recognition techniques, and facilitates a solution to the problem of tracking through conventionally difficult regions containing faulting and other geophysical anomalies, where horizons are discontinuous.

Cellular neural network for seismic horizon picking

From the experimental results in seismic bright spot pattern, the picked horizons can match the visual inspection and improve the seismic interpretation.

Neural networks as an intelligence amplification tool: A review of applications

Artificial neural networks are an intelligence amplification toolkit that allows the interpreter to focus on the important information and couple the speed and efficiency of the computer with the pattern recognition and association capabilities of the brain to aid the exploration process.

A hybrid of neural net and branch and bound techniques for seismic horizon tracking

A hybrid of neural net and branch and bound search techniques for seismic horizon tracking is described in this paper and its results on real as well as synthetic data are shown. Seismic horizon

Toward More Robust Neural-Network First Break And Horizon Pickers

This work shows that significant improvements can be realized by incorporating a variety of multi-trace prediction criteria into the neural network by using a cascade-correlation network whose inherent speed and incremental learning ability make it possible with most data sets to train and apply the network interactively.