High Performance Computing Workshop Hackathon

11 September (Online) & 15 September (On Site)

Building 1, Level 2 Multi-Purpose Room (MPR) 

KAUST, University Campus, Saudi Arabia

Dear Innovators, Data Enthusiasts and Curious Minds

We are thrilled to invite you to an exciting and transformative event at the upcoming High Performance Workshop at KAUST, Saudi Arabia. Join us in this exciting journey to transform the way we handle and interpret seismic data. By developing and utilizing embeddings, we can unlock new potentials in seismic exploration and subsurface analysis. Let your creativity and expertise shine as you tackle these challenges and contribute to the future of seismic data processing. Happy hacking!

Schedule


Wednesday, September 11, 2024

ONLINE via MS Teams / Zoom Link:  Introduction & Information 

  • 09:00am - 12:00pm
    • Welcome Note
    • Learn about Launchpad and compute environment
    • Explore the data and demo notebooks
    • Meet the team and start thinking / working on the project


Sunday, September 15, 2024

ONSITE at Building 1, Level 2 Multi Purpose Room (MPR):  In-person Meeting

  • 09:00am - 09:30am:  Meet & Greet 
  • 09:30am - 12:00pm:  Welcome Note, Projects and Teams Overview
  • 12:00pm - 13:00pm:  Lunch
  • 13:00pm - 16:00pm:  Teams work on their approach & preparing presentations
  • 16:00pm - 17:00pm:  Teams to present their findings and presentations
  • 19:00pm - 21:00pm:  Ice Breaker sponsored by Shell:  Winner Announcement & Awards

Do I need to register for the hackathon?

Participation is included in the workshop registration, NO additional registration is required, however please do let us know if you are attending to ensure you receive all the updates leading up to the event. 

If you will be attending the Hackathon, please inform us by emailing nlr@eage.org 

Embeddings for Pre-Stack Seismic Data

In the realm of seismic exploration, handling vast quantities of pre-stack seismic data presents significant challenges. Traditional methods of seismic data processing often involve extensive computational resources and time-consuming workflows. 

Seismic data, particularly in its pre-stack form, is inherently large and complex. However, the sheer volume of data poses storage, processing, and analysis challenges. Embeddings offer a solution by condensing the data into a manageable size without losing critical details. 

This compression facilitates faster subsurface analysis and paves the way for advanced applications like real-time monitoring and automated interpretation, and end-to-end inversion for subsurface properties using deep learning approaches.

 

Hackathon Goals and Challenges

The hackathon at EAGE HPC Workshop 2024 at KAUST aims to explore the feasibility of building embeddings for pre-stack seismic data and their applications. Participants will fine-tune or train vision, text, or multimodal transformer models to generate embeddings for entire 2D shot-gathers. These embeddings will then be used to enhance various seismic analysis tasks.

Image credit: Metamorworks

Participants are invited to:

Generated Embeddings:

Develop embeddings for 2D shot-gathers arranged in the gather of their choice. These embeddings should be capable of capturing the essential features of the seismic data while significantly reducing its dimensionality.


Cluster and Classify:

Utilize the embeddings to cluster and classify shots based on characteristics such as marine or land data, shallow or deep water environments, synthetic or real data, and the presence of strong ground roll. This task will help in evaluating the discriminative power of the embeddings.


Complement with Other Values:

Integrate additional values or data sources, such as geological or environmental data, to enrich the embeddings and improve the robustness of the analysis.


Image credit: Sadaqat Ali Syed