COURSE OVERVIEW:

Machine Learning Seismic Facies Classification and Validation is an advanced course focused on the automation of seismic pattern recognition. Traditionally, seismic facies analysis is a labor-intensive process prone to human subjectivity. This course demonstrates how supervised and unsupervised machine learning algorithms can be trained to recognize and categorize complex seismic textures, such as salt bodies, channel systems, and carbonate build-ups, with higher speed and objectivity. Participants will explore the transition from manual "picking" to data-driven classification.

The coverage includes the selection and extraction of seismic attributes that serve as the primary features for machine learning models. Attendees will study various classification architectures, including Self-Organizing Maps (SOM), Random Forests, and Convolutional Neural Networks (CNN). A critical component of the course is the validation of these models, ensuring that the machine-derived facies are geologically plausible. This involves calibrating the results with well-logs and core data to bridge the gap between seismic signals and actual lithological units.

Furthermore, the course addresses the challenges of data imbalance and noise in seismic volumes. Participants will learn how to design robust training datasets and implement cross-validation strategies to prevent overfitting. The curriculum emphasizes the interpretability of results, teaching geoscientists how to audit machine learning outputs and integrate them into their structural and stratigraphic workflows. By the end of the course, participants will be able to implement state-of-the-art automated facies classification workflows that significantly reduce interpretation lead times while increasing predictive accuracy.

COURSE OBJECTIVES:

After completion of this course, the participants will be able to:

• Define the technical workflow for machine learning in seismic facies analysis.
• Select and compute optimal seismic attributes for facies discrimination.
• Implement Unsupervised Learning (Clustering) for exploratory facies mapping.
• Apply Supervised Learning for targeted classification of specific geobodies.
• Design high-quality training datasets from interpreted seismic sections.
• Evaluate the performance of classification models using confusion matrices.
• Utilize Convolutional Neural Networks (CNN) for 3D seismic texture recognition.
• Validate machine learning results using blind-well tests and core data.
• Manage data imbalance issues in rare facies detection (e.g., thin beds).
• Perform multi-attribute analysis to improve facies separation.
• Understand the role of "Transfer Learning" in seismic interpretation.
• Integrate automated facies maps into reservoir characterization workflows.
• Mitigate the impact of seismic noise and artifacts on classification models.
• Audit and QC machine learning outputs for geological consistency.

TARGET AUDIENCE:

Geophysicists, Interpretation Geologists, and Data Scientists focused on automating subsurface mapping and reservoir prediction.

TRAINING COURSE METHODOLOGY:

A highly interactive combination of lectures, discussion sessions, and case studies will be employed to maximise the transfer of information, knowledge, and experience. The course will be intensive, practical, and highly interactive. The sessions will start by raising the most relevant questions and motivating everybody to find the right answers. The attendants will also be encouraged to raise more of their questions and to share in developing the right answers using their analysis and experience. There will also be some indoor experiential activities to enhance the learning experience. Course material will be provided in PowerPoint, with necessary animations, learning videos, and general discussions.

The course participants shall be evaluated before, during, and at the end of the course.

COURSE CERTIFICATE:

National Consultant Centre for Training LLC (NCC) will issue an Attendance Certificate to all participants completing a minimum of 80% of the total attendance time requirement.