Tutorial on using convolutional neural networks (CNNs) for automatic interpretation of geothermal temperature and pressure logs. A good tutorial for those in the geosciences (geothermal, water, oil) who are interested in exploring application of CNNs for well log picking. The tutorial includes two accompanying Google Colab notebooks using TensorBoard for visualization as well as several poll questions interspersed in the presentation.
First Google colab tutorial (CNN by hand): https://colab.research.google.com/drive/1cNj09I5YRrokEKu8xJs5P3rQLto3OPKD?usp=sharing
Second Google colab tutorial (CNN with Tensor Flow): https://colab.research.google.com/drive/1F-2cOcw7hM-A7gfUuhAlXgEYi-elI8Lw?usp=sharing
00:00 Intro and overview
02:20 What are convolutional neural networks?
08:00 What is the problem we are looking to solve? Well log interpretation
10:11 Convolution exercise by hand - poll
14:04 Convolutional neural network by hand - Google colab exercise #1
25:28 Part II – training CNNs with Tensor Flow and starting to run colab notebook #2
27:07 Overview process for creating CNN model and CNN architecture elements
42:28 Demo: using TensorBoard
45:15 CNN code with TensorFlow - Google colab exercise #2
