ML in Mineral Discovery Competition (NRCan)

Starts: April 23, 1 pm ET [View this page in French]

🔥

4 week Competition | $4000 in prizes | Guaranteed prizes for all reaching minimal goal | Resources recommended & weekly meetings with Advisors & peers to learn if relatively new!

🤔

This project is meant to help effectively narrow down the search space for the exploration of critical metals across Canada, the United States, and Australia.

💡

General knowledge of ML algorithms is needed for this competition but prior knowledge of Mineral Discovery is optional. (Python is the encouraged language of submission)

Made possible by the Digital Accelerator at NRCan.

Advisors

~ Mohammad Parsa | Technical Advisor Post Doc Fellow in AI Guided Mineral Exploration

~ Christopher Lawley | Domain Advisor Gold Metallogenist at Geological Survey of Canada

Overview

The goal in this competition is to exploit publicly available datasets to develop predictive models for Mississippi Valley-type (MVT) and clastic-dominated (CD) Zn-Pb deposits.

The competition will begin by an introduction to the advisors & potential teammates.

Advisors will share resources & insights to get participants up to speed with the necessary domain & technical knowledge needed to start work on the problem.

Participants must then use their ingenuity to build on that in this 4-week competition.

There will be weekly 1hr meetings with Advisors to help unclog any bottlenecks.

Outside of meetings, the recommended time to invest is between 5-15hrs depending on your prior experience.

Meeting Link: https://meet.jit.si/ML-basedMineralProspectivityMapping

Weekly Meeting Time: 1-2 PM ET on Saturdays (Starting the 23rd April)

Slack channel : click here to join (Gentle nudge to join & shoot an intro!)

Minimal Goals + Prize Distribution

Minimal Goal:

Predictive models with AUC values > 0.9 meet the minimal goal.

The higher the AUC values, the more robust the predictive models.

Don’t worry if you have no clue what’s an ‘AUC value’ - Advisors will provide resources.

Prize Distribution:

$1800 best performing model submissions

$1200 most innovative data processing submission

$1000 distributed among all other teams that reach the minimal goal (ie AUC > 0.9)

Logistics

You can compete alone or in upto a team of 4 people. (Most successful teams are 2/3 people)

You’re welcome to start work alone, then make a team mid-way.

Or you can approach the problem as a team from day 1.

Feel free to join the competition slack channel & shoot a quick intro to start finding a team.

Find the nitty gritty details of the competition (dataset / past attempts etc) here.

Tentative Timeline

#	Major Milestones	Expected time to finish
1	Get familiar with the domain	1 week
2	Data exploration & cleaning	1 week
3	Get to a baseline model	1 week
4	Optimization of model	1 week

Why join - beyond all the other cool reasons?

We host one of the most diverse ML communities in the world. Over the course of this competition:

You’ll get an immersion into that community & walk out with some cool new friends.

All participants reaching the minimal project goal will receive a gift.

You’ll have a chance to learn the foundations of ML-aided Mineral prospectivity mapping.