Artificial Intelligence for Geoscience (Spanish Language Course)

Artificial Intelligence for Geoscience equips participants with the theoretical and practical knowledge needed to apply Machine Learning and Deep Learning concepts to the field of geosciences.

Machine Learning is a subfield of Artificial Intelligence, which is based on trying to imitate the actions of human beings through the training of algorithms. This branch of Data Science is booming in various areas of geosciences, including electrical well log interpretation, reservoir characterization, seismic interpretation, identification of areas with high mining potential, among others.

Deep Learning focuses on the use of neural networks and applying the "back-propagation" method to adjust errors resulting from different iterations, and, when enough data is available, to obtain results far superior to those obtained by "classical" learning algorithms in Machine Learning.

This course will allow participants to apply the knowledge and algorithms learned immediately, both in their research, as well as in their professional career.

Course Structure

SESSION I: PYTHON BASICS (November 7)

• Variables
• Types of Data and Data Management
• Definition and Execution of Functions
• Primary Python Libraries:
  • Numpy
  • SciPy
  • Pandas
  • Matplotlib
  • Seaborn
• Exploratory Data Analysis
• Exercises:
  • Plot a geochemical dataset
  • Visualize, organize, and analyze an oil/gas production dataset

SESSION 2: APPLIED PYTHON (November 14)

• Wavelet (Ricker) in time and frequency
• Well logs
• Display Geospatial Data (Mineral information)
• Seismic volume load
• Post-Stack seismic attributes calculation
• Exercise:
  • Generate a Ormsby, Klander, Butterworth wavelet (Time and Domain)
  • Calculate Coherence Attributes using Python

SESSION 3: MACHINE LEARNING (November 21)

• Supervised vs. Unsupervised
• Regression:
  • Lineal
  • Logistic Regression
• Classification:
  • KNN
  • SVM
• Clustering
  • K- Means
  • Decision Trees
• Dimension Reduction
  • Principal Component Analysis (PCA)
• Exercise:
  • Facies Classification of Well Logs
  • Use Regression Methods to forecast production in an oil well

SESSION 4: DEEP LEARNING (November 28)

• Programming a Neural Network (step by step)
• Application of Neural Networks for facies prediction
• Neural Networks for:
  • Missing Well logs
  • Production forecast of an oil well
• Use CNN for object detection in seismic images
• Exercise:
  • Use CNN to predict salt bodies in a seismic dataset
  • Classify 3D Seismic Facies using Deep Neural Networks