Artificial system

A tool for the future of mining

When studying large areas that are difficult to access, geologists often have to interpret airborne geophysical data. Computer tools are valuable for analyzing aerial images of these sites and generating a preliminary interpretation useful to geologists. Neural networks used in artificial intelligence have also proven to be very effective in analyzing photographs or videos.

The project, funded by the Fonds de recherche du Québec – Nature et technologies (FRQNT), in partnership with the Ministry of Energy and Natural Resources (MERN), as part of its second sustainable mining partnership research program, will analyze all airborne image data to predict the nature of the subsoil. The generated algorithms can be used as guides to target mining exploration sites.

Parallel Neural Networks

This new approach to artificial intelligence aims to train parallel neural networks, each processing a type of variable, such as aeromagnetic or electromagnetic data. Once the processing is complete, a system will combine the predictions from each network to generate a final prediction.

Professor Gloaguen’s team is working with experts in artificial intelligence applied to the mining sector for this network development project, the first phase of which is already underway. The collaborators of the team include professors Bernard Giroux and Peter Simon Ross from INRS, Martin Blouin from Geolearn (an INRS startup), Jean-Philippe Payment of Mira Geoscience, Guy Desharnais of Osisko Gold Royalties and Antoine Caté of SRK. The team will use data from these partners and SIGÉOM, of Quebec geomining information platform.

Postdoctoral researcher Mojtaba Bavand Savadkoohi, in Professor Gloaguen’s team, is currently working on a neural network capable of applying the same resolution to all data. “This will allow us to use satellite, aerial and helicopter images. Getting them in high resolution is essential, because the data is not standardized,” explains Professor Gloaguen.

During the second year of the project, the team will be able to build a deep network architecture and then teach it to recognize geophysical data, a step that will take several months. During the third year, the team will evaluate the generalization potential of the algorithm. “We are going to test the system to see if it can be generalized to all Quebec or whether it will have to undergo a learning phase for each geological region,” explains Professor Gloaguen.

About INRS

INRS is a university dedicated exclusively to research and graduate training. Since its creation in 1969, INRS has played an active role in the economic, social and cultural development of Quebec and ranks first for the intensity of research in Quebec. INRS is made up of four interdisciplinary research and training centers in Quebec, Montreal, Lavaland Varennes, with expertise in strategic sectors: Water Land Environment, Energy Materials Telecommunications, Urbanization Culture Society and Armand-Frappier Health Biotechnology. The INRS community includes more than 1,500 students, postdoctoral fellows, faculty members and staff.


SOURCE National Institute for Scientific Research (INRS)

For further information: Audrey-Maude Vézina, INRS Communications and Public Affairs Department, 418 254-2156, [email protected]