As well as classifying images within documents it is also possible to detect geological features on them. I built a deep learning model to detect salt diapirs and applied it to unseen data as shown in the montage. This can be useful to enhance search & discovery, especially if there is little/no text on or... Continue Reading →

This is what (a subset) of 5,000 labelled subsurface / geoscience images looks like which are typically found in documentation! Happy to share freely with anyone for non-commercial use supporting the geoscience community. These can be used to train a machine vision classifier to help geoscientists sift through this vast amount of information in their... Continue Reading →

Results so far of my deep learning object detection efforts for fossil ammonites. Video here: https://www.linkedin.com/posts/paulhcleverley_geology-fossils-palaeontology-activity-7118884143685349376-3mt4?utm_source=share&utm_medium=member_desktop I wanted a model that could work on my mobile phone to pick out fossils in-situ on the beach, to experiment and see what is possible. More work to do but encouraging signs as a a side hobby!I'm using... Continue Reading →

Topseal performance is critical for geological disposal sites, whether that is carbon capture & storage or radioactive storage, as well as natural hydrogen or oil & gas exploration. I've used patterns in text only using machine automated techniques, to visualise Lithostratigraphic Formations by their association to clues for 'seal' (y-axis), by the overall sentiment of... Continue Reading →

Latest hobby project is detecting fossils (in this case ammonites) from photographs. Labelling examples in existing images, building a deep learning model and applying it to new information. Whether this actually assists me scan the shingle using my mobile phone on my next fossil collecting trip remains to be seen! There is quite a bit... Continue Reading →

Another sentiment visualisation using just text embeddings, tracking changes over time. In this case a subset of North Sea Transition Authority offshore license reports between 2008-2017 for the word vector 'seal'. These types of techniques can support analogues and insights for renewables, carbon capture and storage sites, subsurface radioactive storage, oil & gas exploration, mineral exploration, geohazards... Continue Reading →

Its all about the data. There are some fascinating interactive visualisations avalable from the Organisation for Economic Co-operation and Development (OECD). This chart shows the flow of Venture Capital (VC) investment in data startups by industry, from one country to another in 2022. These can be animated through time, 2023 showing growth in healthcare and... Continue Reading →

The area of Natural Language processing (NLP) research has exploded in recent times. Building Large Language Models (LLM) is a big player within the NLP landscape, but not the only game in town. I would like to point you towards an excellent paper by Schopf et al (2023) who classified and analysed NLP research papers... Continue Reading →

Mathur et al (2023) published an interesting paper recently. Transformers for Site Assessment for Carbon Capture and Sequestration using Legacy Well Data Y Mathur, J Chen, I Folmar, Z Dong, Q Su, L Lu, M Sidahmed Third EAGE Digitalization Conference and Exhibition 2023 (1), 1-5, 2023 Carbon Capture and Sequestration (CCS) is one of the... Continue Reading →

I've been experimenting using text embeddings to generate sentiment of a corpus of documents. In this approach it is generated by geological age (but can be other contexts). Taking any input query e.g. "aquifer" then combining that (adding vectors) with geological age vectors and comparing to the cosine of the vector of various sentiment themes,... Continue Reading →