Partnership between Total and Google in Artificial Intelligence

iero (42)in #ai • 7 years ago

Google Cloud et Total Exploration Production partner to develop Artificial Intelligence solutions for Geoscience and subsurface data analysis.

As announced in the press release and in the video of Jean-Michel Lavergne, Senior Vice President, Strategy, Business Development, R&D at Total E&P, Total will locate a joint team of Geoscientists in Google's offices of Sunnyvale, California to develop, with Google Cloud researchers and engineers, artificial intelligence solutions around two streams:

computer vision : study of seismic images
natural language processing : analysis of technical documents

Why this collaboration is important for Total

At the beginning of a new study, Geoscientists (Geologists, Geophysicists, Reservoir engineers and Geoinformation) gather data, from recordings (seismic images, well logs) and technical documents (study reports, synthesis). Then, they analyze these data in order to understand the subsurface and find the signals leading to a discovery and to an optimal development.

After decades of studies, the amount of available data is very important. Vintage data might be refuted by recents discoveries or by updated interpretations with new techniques and concepts.

Extensive evaluation takes a lot of time and effort. Programs from the two streams will enable Geoscientists to get rapid thoroughful analysis of their data so that they can focus on added-value steps of new subsurface studies.

Total's research effort, through integrated teams of researchers, data scientists, developers and geoscientists, is already available trough advanced tools like Sismage. Total wants now to build on the latest advances in machine learning and offer even more innovative services to its teams.

To achieve these goals and deploy them quickly, Total has decided to partner with a major artificial intelligence and cloud player.

Jamboard

Why Google

Google is the leader in artificial intelligence. Their tools are used every day by millions of people, and they continue to regularly publish new services directly from their research. For example:

Tensorflow, the most widely used artificial intelligence framework in the world, Opensource since 2015
Inception, the famous neural network architecture known since 2014 for its performance in image recognition
The Tensor Processing Unit (TPU), the best hardware architecture available for deep learning
AutoML, Google's first consumer tool allowing companies to easily label their images and re-train (via TPU transfer learning) models on new data.

Google Cloud offers a powerful, integrated cloud infrastructure with all services needed for Total to develop and share results with operational teams very quickly.

An innovative collaboration

For the first time, a team composed of complementary skills, will work in an integrated way to apply the latest advances in deep learning to Geoscience use-cases. This collaboration will be set up within the Advanced Solution Laboratories of Google Cloud.

This unique partnership between a Cloud giant and a Energy major, lays the fundation of an ambitious program which has the potential to transform the way Geoscientists works. The first steps will be to consolidate the available information, making it more accessible, providing new tools and new ways of working for Geoscientists in Total.