Semi-supervised Impedance Inversion by Bayesian Neural Network Based on 2-d CNN Pre-training

TL;DR

This paper introduces a semi-supervised seismic impedance inversion method using a 2D CNN and Bayesian inference to improve accuracy and estimate uncertainty, validated on Marmousi2 and SEAM models.

Contribution

It enhances semi-supervised impedance inversion by replacing 1D CNNs with 2D CNNs and incorporating Bayesian inference for uncertainty estimation.

Findings

Improved prediction accuracy with 2D CNN architecture.

Effective uncertainty estimation via Bayesian framework.

Validated approach on Marmousi2 and SEAM models.

Abstract

Seismic impedance inversion can be performed with a semi-supervised learning algorithm, which only needs a few logs as labels and is less likely to get overfitted. However, classical semi-supervised learning algorithm usually leads to artifacts on the predicted impedance image. In this artical, we improve the semi-supervised learning from two aspects. First, by replacing 1-d convolutional neural network (CNN) layers in deep learning structure with 2-d CNN layers and 2-d maxpooling layers, the prediction accuracy is improved. Second, prediction uncertainty can also be estimated by embedding the network into a Bayesian inference framework. Local reparameterization trick is used during forward propagation of the network to reduce sampling cost. Tests with Marmousi2 model and SEAM model validate the feasibility of the proposed strategy.