"Geng et al., 2020, Deep learning for relative geologic time and seismic horizons" is the title of the paper that provides a framework for calculating 2D RGTs and extracting horizons from them. Since extracting true RGT from field data is not straightforward, the network is trained using a synthetically-generated dataset. The proposed solution is tested on a range of field data that demonstrate promising results.
I am particularly interested in such development. I believe that the future of structural interpretation would be tight around with Deep Learning-based RGT, meaning we would extract faults, horizont, and other features directly from RGT;
The link for the paper: https://www.researchgate.net/publication/340317169_Deep_learning_for_relative_geologic_time_and_seismic_horizons
The link for the LinkedIn post