Page 264 - Machine Learning for Subsurface Characterization

228 Machine learning for subsurface characterization

FIG. 8.2 Comparisons of measured NMR T2 distributions against those synthesized by processing
inverted logs in the testing dataset using (A) OLS, (B) LASSO, (C) ElasticNet, (D) SVR, (E) kNNR,
and (F) ANN models.

4 Deep learning models
Most deep learning models are based on deep neural network architecture
comprising several hidden layers. There is no clear boundary between
shallow- and deep learning models. The definition of deep learning evolves
with the development computation speed and data size. Unlike shallow-
learning models, deep learning models can be designed to have specific
architecture that facilitates better approximation, abstraction, and
generalization of information in training dataset. In this chapter, we apply
four distinct deep learning models for the desired NMR T2 synthesis. Three
of the four deep models first learn features and abstractions essential for