Deep Learning and artificial intelligence can play a powerful role in exploration and mining. Achieving good results depends upon the completeness and correctness of the input data. This is especially the case with location within data sets. Data scientists, geoscientists, and mine engineers must collaborate to correctly transform the data and thus achieve meaningful results.
Data Analytics for Digital Marketing Lecture for Advanced Digital & Social Me...
A Miners Drift Volume 01 Issue 02 2018-APR-16
1. A Miner’s Drift
A Journal of Occasional Explorations
Joseph Starwood – Digital Advisor, Geologist, & Geophysicist
Volume 01
Issue 02
2018-APR-16
Deep Learning &
Artificial Intelligence
in Mining
The importance of knowing where you are
2. Deep Learning & Artificial
Intelligence in Mining
The importance of knowing where you are
Stories of lost mines abound. Each year in the United States, over 8000
people set out to find the legendary Lost Dutchman Gold Mine; most
guided by only cryptic maps. And, each year, the mine goes
undiscovered. Precise location is critical to exploration and mining.
There is no room for error – let alone cryptic. This hold true when
applying deep learning and artificial intelligence methods; where we
seek subtle signatures in the data.
3. 2018-APR-16 A Miner’s Drift Volume 01 Issue 02 Copyright 2018 All rights reserved
“There’s gold in them thar hills”
__Mulberry Sellers, a character by MarkTwain
The problem, as always, is where
exactly that gold is located in those
hills. Or, for mining in general, where
exactly the ore is located.
Location is fundamental to mining. It
is critical that a miner know where
the ore body and its boundaries are
located. Miners must also know their
own position with respect to the ore
body as well as with respect to
property boundaries and other
important reference points.
The same holds true for deep
learning and artificial intelligence
models. Correct locations for input
data points will yield useful results.
However, incorrect locations will lead
to erroneous results.
Determining the correct spatial
location of any given data point is not
as easy as one might think. Mining is
a peculiar business. The various data
sets forming the input data for the
models may not, and often do not,
share the same basis for location. In
other words, the location values for
one data set may not directly
correspond to those in another data
set.
Geoscientific data, in particular, relies
on correctly locating data points in
three dimensional space. This poses
complex challenges for data
scientists.
Let’s look at some of the challenges
data scientists must overcome if their
models are to be correct:
INCH BY INCH,YARD BYYARD
Measurement of length is one of the
most basic ways in which input data
sets may differ from one another.
The length measurements may
utilize the imperial system (i.e.: inch,
foot, yard, etc.) or the metric system
(i.e.: meter, kilometer, etc.).
This happens when one work activity
used the imperial system and a
another work active used the metric
system. This case might arise when a
mining company acquires an existing
property. The earlier pre-acquisition
work relied on one measurement
system, and the acquiring mining
company uses a different system.
Mixing data based on the imperial
system with data based on the metric
system will yield erroneous results.
“1065.5” in feet is not the same as
“1065.5” in meters. The imperial
measurement location (1365.5,
1241.3, 515.8) is not the same as the
metric measurement location
(1365.5, 1241.3, 515.8).
PAGE 3
4. 2018-APR-16 A Miner’s Drift Volume 01 Issue 02 Copyright 2018 All rights reserved
When different measurement
systems are used across the various
input data sets, the data first must be
converted to use just one preferred
measurement system.
TIP: Always verify the measurement
system for each input data set before its
use in deep learning or artificial
intelligence modeling.
THE MAP IS NOTTHETERRITORY
In mining, geoscientific data may be
presented on maps and cross-
sections. When presented this way,
three dimensional data is reduced to
a two dimensional view. Thus, the
third dimension is omitted.
For example, an exploration map of
iron (Fe) may show several data
points with their values along with
contour lines inferring the boundaries
for different iron concentrations.
However, the map may well ignore
the local topographic relief; the
vertical coordinate). The data is
presented only in terms of its X andY
(North-South and East-West)
positions
Similarly, a cross-section (a vertical
map) highlighting silver (Ag) at the
mine may show several data points
with their values along with contour
lines inferring the boundaries for
different silver concentrations
A cross-section may or may not be
aligned with either the North-South
or East-West axes.This means that
we may know the vertical
measurement (Z-axis), but not the
horizontal measurements (X- andY-
axes).
To complicate things further, the
cross-section may have one or more
‘dog-legs’. This means that the cross-
section, is composed of two or more
contiguous non-coplanar rectangles.
When viewed in map perspective,
such a cross-section has two or more
non-collinear segments.
TIP: Remember that maps and cross-
sections are two dimensional
representations of data; not the data
itself. Always locate and use the original
data for deep learning or artificial
intelligence modeling.
PAGE 4
5. 2018-APR-16 A Miner’s Drift Volume 01 Issue 02 Copyright 2018 All rights reserved
GEOGRAPHICCOORDINATE
SYSTEMS
A geographic coordinate system is
used to specify locations on Earth.
These locations are expressed using
numbers, letters, and symbols. Each
location represents a position in a
three dimensional space with two
numbers representing the horizontal
position and one number the vertical
position.
Three coordinate systems are used
most often in mining: 1) Latitude,
Longitude, & Elevation; 2) Universal
Transverse Mercator & Elevation; and
3) Mine Grid & Elevation.
When different measurement
systems are used across the various
input data sets, the data first must be
converted to use just one preferred
coordinate system.
LATITUDE & LONGITUDE
The Latitude & Longitude system for
horizontal position is perhaps the
most familiar. Longitude is measured
starting from a designated north-
south line called the Prime Meridian.
Latitude is measured starting from a
designated east-west line called the
Equator. In both, measurements are
expressed as angles using degrees,
minutes, and seconds of arc. Vertical
position is often measured with
respect to mean sea level.
UNIVERSAL TRANSVERSE MERCATOR
The UniversalTransverse Mercator
system is a bit more complicated. It
uses a pattern of rectangular regions
known as zones. Locations are
specified as easting and northing
planar coordinate pairs in their zone,
and are measured in meters. Again,
vertical position is often measured
with respect to mean sea level.
PAGE 5
6. 2018-APR-16 A Miner’s Drift Volume 01 Issue 02 Copyright 2018 All rights reserved
MINE GRID
The Mine Grid system establishes a
Cartesian coordinate system based
on a local reference point. The top-
bottom axis (grid north-south axis)
may or may not be aligned to true
North. Locations are measured as
east or west and as north and south
of the reference point, and are
expressed in imperial or metric units.
The Mine Grid system is unique in its
treatment of vertical position. The
designated local reference point
serves as the origin for measuring
other locations as above or below.
However, this origin may be given an
arbitrary value other than mean sea
level and perhaps other than zero.
Mind Grids are often oriented to the
strike of the ore body or another
dominate geographic feature. They
may also be referred to as ‘Local
Grids’ or ‘Local Coordinates’.
TIP: Always verify the type of
coordinate system for each input data
set before its use in deep learning or
artificial intelligence modeling.
TRUE NORTH
Across much of the Earth,True North
differs from Magnetic North. This
arises because the Earth’s magnetic
pole is offset from the Earth’s
rotational (axial), orTrue North, pole.
The difference betweenTrue North
and Magnetic North is referred to as
the magnetic declination. A given
place on Earth, say the exploration
site or the mine site, will have a
specific magnetic declination.
Location measurements must be
corrected toTrue North.
Ideally, geoscientists, engineers, and
others will have applied corrections
to the location data. However, this
may not always be the case. Some
compasses do not allow the user to
apply a correction for the
magnetic declination.
TIP: Always verify the
coordinate system used for
each input data set before its
use in deep learning or artificial
intelligence modeling.
PAGE 6
True
North
7. CONCLUSION
Deep Learning and artificial intelligence can play a powerful role in
exploration and mining. Achieving good results depends upon the
completeness and correctness of the input data. For location data, the raw
data must first be corrected to ensure that it uses the preferred units of
measure and geographic coordinates, and that it is properly corrected to
True North. Doing so prepares the input data for deep learning and artificial
intelligence methods. Data scientists, geoscientists, and mine engineers
must collaborate to correctly transform the data and thus achieve
meaningful results.
8. During his early career, Joseph worked as an exploration geologist and
geophysicist. He focused on gold, silver, copper, and molybdenum deposits in
the western United States. His work included Kuroko style gold systems. He
also worked in various capacities for coal mining, oil & gas, and heavy
construction companies.
Joseph has presented at:
▪ InfraGardGreat Lakes Regional Conference – 2014
▪ Society for Mining, Metallurgy, and Exploration Annual Meeting – 2014
▪ Great Lakes Software ExcellenceConference (GLSEC) – 2011 & 2013
▪ Calvin College Colloquium Series – 2010
▪ Integrate: Uniting theWorld of IT Symposium – 2010
▪ SEI ArchitectureTechnology User Network (SATURN) – 2009
▪ Financial Services Solutions Symposium (FSSS) – 2008 & 2009
He also contributed to:
• Software Engineering Institute (SEI) – Hard Problems in SOA Workshop – 2008
Contact theAuthor:
• LinkedIn: www.linkedin.com/in/josephstarwood
• e-Mail: JosephStarwood@JosephStarwood.com
Known within Microsoft Corporation as 'The Mining
Guy', Joseph Starwood is a Digital Advisor who helps
mining companies, equipment suppliers, and service
providers extract value from their information
technology (IT) and operational technology (OT)
investments and assets. He co-authored the Mining
Book of Dreams as well as the Microsoft Upstream
Reference Architecture for Mining (MURA-Mining).
PLEASE NOTE
The author is an employee of Microsoft Corporation. The views, thoughts, and opinions expressed in this text belong wholly to the author, and do not necessarily reflect those of anyone else (the author’semployer, or any other group or individual).
All product names, logos, and brands are property of their respective owners. All company, product and service names used in this text are for identification purposes only. Use of these names, logos, and brands does not imply endorsement.
The information in this text is provided without representations or warranties, express or implied.