What is it about?
This work demonstrates the applicability of a hybrid GADAM (genetic-evolutionary adaptive moment estimation) optimizer for acoustic impedance inversion in a deep learning framework. The seismic impedance computed using the GADAM shows better contrast and higher accuracy in the visualization of the seismic reflectors than the model-based approach and extensively used ADAM (adaptive moment estimation) optimizer.
Photo by Markus Spiske on Unsplash
Why is it important?
This study proposes an application of a hybrid GADAM optimizer for achieving global optimization in seismic impedance inversion. Being a hybrid optimizer, GADAM integrates the advantages of ADAM and GA into a single optimization algorithm that enhances the probability of attaining a globally optimal solution/optimal impedance model by avoiding the local optima. We found that this novel approach to estimating the seismic impedance improves the overall visualization and correlation, which implies that the solution obtained from the GADAM optimizer has better generalization performance in contrast to ADAM.
Read the Original
This page is a summary of: Genetic-evolutionary adaptive moment estimation-based semisupervised deep sequential convolution network for seismic impedance inversion: Application and uncertainty analysis, Geophysics, March 2023, Society of Exploration Geophysicists, DOI: 10.1190/geo2022-0061.1.
You can read the full text:
The following have contributed to this page