This course aims at introducing the basics and fundamentals of deep learning (DL) and illustrating its values in assisting seismic interpretation. By completing this course, participants will (1) learn the basic DL concepts and principles, (2) get familiar with the DL pipelines for common seismic interpretation tasks through case studies, and (3) build their own DL algorithms through hand-on examples.
Duration
8 hours
Intended Audience
Entry Level
Prerequisites (Knowledge/Experience/Education required)
The course is designed to be followed by anyone with a basic knowledge of math and physics. However, experience in seismic interpretation, machine learning or python programming will be useful. The course is an ideal opportunity for participants who have ongoing projects that DL could potentially bring value to.
Why Attend?
Deep learning, since its first introduction from image processing domain to seismic interpretation domain around 2016, has proven its great value in assisting various interpretation tasks and is considered essential in improving existing and/or developing new interpretation workflows and algorithms. Such efforts won’t be successful without a solid understanding about what the DL basics are and more importantly how it should be implemented and applied to seismic data. This course is to pave the foundations, bridge the gaps between DL and seismic, and stimulate more innovations in DL-based seismic interpretation.
Course Description
This course offers an introduction on how to develop a deep learning solution to seismic interpretation tasks. Therefore, the course will develop participants’ expertise in integrating DL with seismic data from the following sub-topics.
- Introduction to deep learning: Understanding the basic concept and principles of deep learning is vital to any successful solution development and implementation. The course will start with introducing the fundamentals of deep learning, including neural network, loss function, optimization, etc.
- Case studies: One good example usually tells more than pages of descriptions. The second part of the course will collect and go through some of the best examples on how deep learning helps resolve a seismic interpretation challenge.
- Hand-on exercise: One major objective of this course is to allow the participants practicing their skills in building a deep learning solution. Therefore, this course will provide example tasks with data and label, so that each participant could program his/her own solution and build a solid foundation on more innovations afterwards.
Learner Outcomes
The participant will:
- Capture the basics of deep learning concepts and principles
- Understand the general pipeline of a deep learning solution through case studies
- Learn how to build a neural network from scratch
- Practice programming to solve a provided seismic interpretation task
