Neural Networks are used for recognizing patterns, for extracting seismic objects and for quantitatively predicting rock properties from seismic inversion products.
The Neural Networks plugin in OpendTect supports supervised- and unsupervised learning approaches for clustering data and for finding non-linear relationships.
For optimal results the Neural Network plugin should be combined with the Dip Steering plugin, which supports pre-processing filters and extraction of attributes along seismic reflectors.
"OpendTect provides state-of-the-art pattern recognition tools to complement your eyeball quantitatively."
--Leon Thomsen, Principal, Delta Geophysics, former Senior Advisor and Principal Geophysicist, BP, and former S.E.G. President
All commercial plugins require OpendTect Pro.
The Neural Networks plugin supports Multi-Layer-Perceptrons (MLP) and Unsupervised Vector Quantizers (UVQ) networks to perform the following tasks:
- Create Object “Probability” Cubes such as Chimney Cube, Salt Cube, Fault Cube, …
- Create maps, or volumes with supervised classification results e.g. revealing fluid content classes, lithology classes, seismic facies classes, …
- Create seismic facies pattern maps through unsupervised waveform clustering
- Create volumes with 3D clusters through unsupervised attribute clustering
- Create rock property volumes such as porosity, Sw, lithology class, … from seismic inversion products such as AI, EEI, Angle Stacks, ...