Post Doctoral Fellow, Mechanical Engineering (Machine Learning in Mining)

Primary Purpose : The successful candidate will be responsible for development of a Machine Learning algorithm to quantify the safety of underground mine roofs and sidewalls using audio recordings of tool impacts on the surface. This is a project in the laboratory of Prof. Travis Wiens in the Department of Mechanical Engineering and is a collaboration with Nutrien (the world’s largest potash miner) and funded by NSERC.

Nature of Work : The PDF will report to Prof. Wiens and work under his supervision. A dataset of audio recordings of tool impacts on mine walls and roofs already exists (and is being expanded by another student), with labels of “drummy” (unsafe) or “tight” (safe) as identified by experienced mine personal. The PDF will be responsible for applying the latest developments in machine learning to develop and optimize the performance of an algorithm to predict whether a recording is drummy or tight. They will also expand it to quantify the level of drumminess and the confidence in this prediction. The dataset is characterized by a small number of samples relative the length of the recording, label uncertainty and moderate signal-to-noise level.

The PDF will be expected to work independently to identify, develop and validate new ideas as well as take the lead in communicating the results via publications and conference presentation.

Although not a formal supervisor, the PDF will have a mentorship and team member role in interactions with a student who is currently taking the recordings (as well as other students working on other projects).

The work environment will be a typical office although visits to an underground potash mine site may be desirable to understand the background and possibly to take new data or improve the data recording process. Hours will be flexible, based around regular university hours and working from home part-time is possible.

Accountabilities : The PDF will be expected to produce regular progress reports, both to their supervisor as well as industrial partners and funding agencies. They will be expected to produce papers for publication in reputable peer-reviewed journals and conferences.

Education : Must have completed a in a related field within the last five years.

Licenses : n / a

Experience :

Required Experience :

• Demonstrated experience in developing machine learning algorithms with small datasets of arbitrary and / or time-series data
• Demonstrated experience with dataset feature engineering and dimensionality reduction
• Demonstrated ability to take the lead role in successfully published paper(s) in high quality journals or peer-reviewed conferences

Optional but Desirable Experience :

Skills : Required and desirable skills related to the above experiences, as well as excellent communication, organizational and interpersonal skills.