216
When the algorithm was run against a test dataset, a success score of over 98% was
achieved. MICROMINE plans to commercialize this machine learning technology. The
vision is to develop technology that transforms the core tray photos into continuous drill
core images for each hole (Figure 6).
MICROMINE currently has technology that can do this. It can join and present 2-dimensional core images on 3-dimensional drill hole strings. Using the technological developments
described in this paper, the manual imaging clipping and labeling processes will be replaced
by ones controlled by machine learning.
The accuracy of automatically positioned clipped core images is unknown. Determining
this accuracy is a subject for later investigations.
5 CONCLUSION
The development of machine learning and the ability to use pre-trained algorithms allows the
rapid development and deployment of this technology to drill hole data. The technology has
already been successfully tested in a project that separated the core and non-core components
in core tray imagery. But similar image processing techniques could be developed to recognize features such as colour and geological structures. Isolating this data and storing it in
databases along with related observations and measures will deliver value to the geologist.
Machine learning can also be applied to non-image data to make geological predictions
during drilling and to check the quality and consistency of manual and electronically collected data. This includes the use of geophysical and geochemical signatures to help in the
isolation and identification of lithological units.
Machine learning is becoming an important tool to the geologist and will improve the
quality and consistency of the data available to processes within the mining chain of activities. If the technology can be effectively applied, the efficiency gains will improve mine profitability through reduced cost, time and increased data quality.
REFERENCES
Bengio, Yoshua; LeCun, Yann; Hinton, Geoffrey 2015. “Deep Learning”. Nature. 521 (7553): 436–444.
Bibcode:2015 Natur.521.436 L. doi:10.1038/nature14539. PMID 26017442.
Bishop, C.M. 2006, Pattern Recognition and Machine Learning, Springer, ISBN 978-0-387-31073-2.
Chollet, F. 2017. “Deep Learning with Python”. O’Reilly Media.
Coolen, A.C.C, R. Kuehn, P. Sollich 2005. “Theory of Neural Information Processing Systems”. Oxford
University Press.
Dimitri P. Bertsekas 2012 Dynamic Programming and Optimal Control: Approximate Dynamic Programming, Vol.II”, Athena Scientific.
Dimitri P. Bertsekas and John N. Tsitsiklis 1996 Neuro-Dynamic Programming, Athena Scientific.
Geron, A. 2017 “Hands-On Machine Learning with Scikit-Learn and TensorFlow”. O’Reilly Media,
Inc, USA.
Krizhevsky, A., Sutsskever, I., Hinton, G. 2012. “ImageNet Classification with Deep Convolutional
Neural Networks”. https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf.
Mohri, Mehryar; Rostamizadeh, Afshin; Talwalkar, Ameet 2012. Foundations of Machine Learning.
The MIT Press. ISBN 9780262018258.
Russell, Stuart J.; Norvig, Peter 2010. Artificial Intelligence: A Modern Approach (Third ed.). Prentice
Hall. ISBN 9780136042594.
Poole, David; Mackworth, Alan; Goebel, Randy 1998. Computational Intelligence: A Logical Approach.
New York: Oxford University Press. ISBN 978-0-19-510270-3.
Yann LeCun, Yoshua Bengio & Geoffrey Hinton Nature 521, 436–444 (28 May 2015). doi:10.1038/
nature14539.
When the algorithm was run against a test dataset, a success score of over 98% was
achieved. MICROMINE plans to commercialize this machine learning technology. The
vision is to develop technology that transforms the core tray photos into continuous drill
core images for each hole (Figure 6).
MICROMINE currently has technology that can do this. It can join and present 2-dimensional core images on 3-dimensional drill hole strings. Using the technological developments
described in this paper, the manual imaging clipping and labeling processes will be replaced
by ones controlled by machine learning.
The accuracy of automatically positioned clipped core images is unknown. Determining
this accuracy is a subject for later investigations.
5 CONCLUSION
The development of machine learning and the ability to use pre-trained algorithms allows the
rapid development and deployment of this technology to drill hole data. The technology has
already been successfully tested in a project that separated the core and non-core components
in core tray imagery. But similar image processing techniques could be developed to recognize features such as colour and geological structures. Isolating this data and storing it in
databases along with related observations and measures will deliver value to the geologist.
Machine learning can also be applied to non-image data to make geological predictions
during drilling and to check the quality and consistency of manual and electronically collected data. This includes the use of geophysical and geochemical signatures to help in the
isolation and identification of lithological units.
Machine learning is becoming an important tool to the geologist and will improve the
quality and consistency of the data available to processes within the mining chain of activities. If the technology can be effectively applied, the efficiency gains will improve mine profitability through reduced cost, time and increased data quality.
REFERENCES
Bengio, Yoshua; LeCun, Yann; Hinton, Geoffrey 2015. “Deep Learning”. Nature. 521 (7553): 436–444.
Bibcode:2015 Natur.521.436 L. doi:10.1038/nature14539. PMID 26017442.
Bishop, C.M. 2006, Pattern Recognition and Machine Learning, Springer, ISBN 978-0-387-31073-2.
Chollet, F. 2017. “Deep Learning with Python”. O’Reilly Media.
Coolen, A.C.C, R. Kuehn, P. Sollich 2005. “Theory of Neural Information Processing Systems”. Oxford
University Press.
Dimitri P. Bertsekas 2012 Dynamic Programming and Optimal Control: Approximate Dynamic Programming, Vol.II”, Athena Scientific.
Dimitri P. Bertsekas and John N. Tsitsiklis 1996 Neuro-Dynamic Programming, Athena Scientific.
Geron, A. 2017 “Hands-On Machine Learning with Scikit-Learn and TensorFlow”. O’Reilly Media,
Inc, USA.
Krizhevsky, A., Sutsskever, I., Hinton, G. 2012. “ImageNet Classification with Deep Convolutional
Neural Networks”. https://papers.nips.cc/paper/4824-imagenet-classification-with-deep-convolutional-neural-networks.pdf.
Mohri, Mehryar; Rostamizadeh, Afshin; Talwalkar, Ameet 2012. Foundations of Machine Learning.
The MIT Press. ISBN 9780262018258.
Russell, Stuart J.; Norvig, Peter 2010. Artificial Intelligence: A Modern Approach (Third ed.). Prentice
Hall. ISBN 9780136042594.
Poole, David; Mackworth, Alan; Goebel, Randy 1998. Computational Intelligence: A Logical Approach.
New York: Oxford University Press. ISBN 978-0-19-510270-3.
Yann LeCun, Yoshua Bengio & Geoffrey Hinton Nature 521, 436–444 (28 May 2015). doi:10.1038/
nature14539.