Reservoir Computing with Machine Learning: Theory and Practice

Workshop 6: Sunday, 2 June 
Lecture Room:9
Conveners: Vasily Demyanov (Heriot-Watt University)
Abdulrahman Moqbel (Saudi Aramco)

Description:

The workshop aims to provide an overview of the state-of-the-art neural computing and its applicability to reservoir applications. The workshop will combine a theoretical introduction to the concept of learning from data and AI applicability in geoscience context; typical examples of AI applications to reservoir modelling problems; and a practical hand-on exercise illustrating machine learning capability with real reservoir data. The workshop will provides a set up for a round discussion of the opportunities provided by computer based learning to tackle difficult data and uncertainty rich reservoir problems. 

The workshop participants will:

  • Become familiar with machine learning concepts.
  • Gain understanding of the variety of learning-based algorithms.
  • Get an illustrative insights from the industry and academia of machine leaning use in reservoir applications.
  • Get a hands-on in tackling different reservoir problems with machine learning methods using real field data.
  • Discuss appropriate use of machine learning and its practical aspects in reservoir studies.


Workshop Programme

09:00Introductory Overview 
V. Demyanov (Heriot-Watt University)
Keynote Introductions to AI Methods in Geoscience and Decision-making:
09:10Introduction to Neural GeoSpatial Data Processing
M. Kanevski (University of Lausanne)
10:00Value of Information and Learning Theory - Geometry of optimal decision-making control
R. Belavkin (Middlesex University)
10:50Coffee break
Shared practice of AI in Reservoir Computing:
11:10Applied Intellectual Systems in Real O&G Practice - From concept to real process implementation
B. Beloserov (Gaspromneft)
11:35Machine Learning and Geophysical Characterization
S. Cersosimo  (Galp)
12:00Machine Learning and Cloud Computing in Linking Reservoir Characterisation and Reservoir Dynamics through Realistic Hierarchical Model Update
T. Buckle & R. Hutton (Heriot-Watt University)
Wrap up round
12:30Lunch break
13:30Hands-on practical ML application paper exercises for decision-making with real reservoir data. 
How the variable and the data selection may impact a learning model outcome? 
How to tune a learning-based model?
How to choose the most appropriate learning model?

16:00Discussion
16:30End of Workshop

Main Sponsors

                   

© EAGE 2018 (version 1.0.5.0)     Privacy     FAQ