4th April 2019
Machine Learning has seen explosive growth in recent years and its methods of learning from data are already widely applied across many areas such as life sciences, finance and social media. Driven by rapid developments in learning algorithms, advances in computing infrastructure and increasing amounts of data from which to learn, machine learning is also starting to play an increasingly significant part in hydrocarbon exploration and production. Machine learning offers the promise of increased efficiency in areas like seismic interpretation and petrophysical analysis, and in understanding the avalanche of data produced by various digital sensors. It also holds the potential to analyse and extract value from the vast amounts of legacy data that through lack of resources are currently left to gather dust.
The machine learning special interest group welcomes all with an interest in the field from those who just want to learn a bit about the technology through to those who have experiences to share. We think it is important for us all, as users of this rapidly evolving technology, to able to ask the right questions about the advantages and limitations of the methods we may increasingly come to rely on. So join this group for education and critical thinking about machine learning as well as networking in an informal atmosphere.
This event is open to PESGB members only. Please note all members whom attend will automatically be registered for the Machine Learning SIG. If you would like to be part of the Machine Learning SIG, but are unable to attend the event please contact firstname.lastname@example.org
Spaces are limited, so please remember to register
Machine Learning for Well Log Quality Control and Reconstruction with David Psaila
In this talk we look at some applications of machine learning to the challenges presented by large well log data sets in a regional seismic reservoir characterisation study. In this setting, issues of data quality caused by variations in factors such as data vintage, hole conditions, logging tools, and geology, require detailed attention when attempting to reconcile well and seismic data. The objective of this work is to apply machine learning algorithms to eliminate time consuming and repetitive tasks such as manual log editing and rock physics modelling. The machine learning results are benchmarked against the work of a team of expert petrophysicists and compare very favourably. Seismic well-ties based on the machine learning results also confirm the quality of the results.
David Psaila is Director of Analytic Signal Limited, a London-based consultancy founded in 2016 to explore the application of machine learning in the upstream oil and gas industry. He has over 20 years experience working for oilfield service companies, developing technology in the areas of reservoir geophysics, geostatistics, reservoir modelling, and exploratory data analysis. David has experience of studies including quantitative time-lapse interpretation, pore pressure prediction, and stochastic seismic inversion; as well as experience in commercial software development on products including Petrel. He holds a degree in Geology from Oxford University and a PhD in Geophysics and Geostatistics from Leeds University.