Halliburton and AAPG are collaborating to bring to AAPG community a Hybrid Physics-based deep neural network (DNN) Hackathon.

The objective of this hackathon, designed for domain geoscientists and engineers, is to hack the provided Python® code and develop DNN models for time series predictions of reservoir behavior.

After we sell out, please sign up on the waiting list -- we will open more seats.

Hybrid Physics/DNN Integration

The use of deep learning models in oil and gas is on the rise.

Deep learning models while showing a lot of promise have limitations when applied to oil and gas problems.

Specifically these limitations deal with being able to incorporate geoscientists’ understanding of subsurface physics into deep learning models.

Subsurface physics can be incorporated in one of many ways a subset being:

Augmentation of training datasets for deep learning models using data generated by physics driven simulators, including physics-based models as a component in an ensemble of data driven deep learning models

Formally incorporating domain physics within the deep neural network structure

Scope and Prerequisites

Domain meets DNN hackathon is designed for practitioners, data scientists, developers, decision-makers and includes hands-on Python coding experiments. OpenEarth™ Community (OEC) , a cloud-based environment will be provided. For this hackathon, participants will focus on

• Augmenting provisioned training dataset using a relevant physics based model
• Modifying and applying supplied algorithm to their own model

Required Bootcamp Webinar July 16. All participants must participate in the pre-event familiarization boot-camp webinar. The bootcamp will orient participants with Python-coding and OEC environment. You will receive log-in information and time.

Agenda

Thank you to our sponsors

Hackathon Host

Coffee Break Sponsor

Patron Sponsor

Patron Sponsor