Development of machine learning algorithms for wearable devices - BC-393

Preferred Disciplines: Computer Science, Data Science, Machine Learning (Masters, PhD, Post-Doc)
Project length: 16-24 months (4 units)
Approx. start date: As soon as possible
Location: Vancouver, BC
No. of Positions: 1
Preferences: None
Company: Form Athletica

About Company:

Form Athletica is building a new revolution in augmented reality for sports. Run by a team of industry veterans with a reputation for creating amazing, industry-first products. This includes part of the team that introduced the world’s first consumer smart eyewear in 2010. Form is headquartered in Vancouver, BC. 

Summary of Project:

Form Athletica is building a new revolution in augmented reality for sports, and we’re looking for a Data Scientist to join our rapidly growing team. In our system, sensor signals from the wearable device are analyzed to give real-time feedback to the user. This role is to expand its capabilities with research into "edge machine learning" -- machine learning prediction designed to run on embedded devices, and how this can complement big data techniques.

We need to compare the performance of different machine algorithms including deep learning on a labelled data set.  Approaches could involve other signal processing and related techniques to improve accuracy of predictions. Work would focus on enabling on-line training to individualize performance for each user.

Research Objectives/Sub-Objectives:

  • Design and implement methods to  improve prediction accuracy of machine learning algorithms
  • Optimize machine learning algorithms for low-power devices

Methodology:

    • Investigate both online and postprocessing approaches, with individual and aggregated data, to improve performance and accuracy
    • Use statistical approaches to minimize the number of false positive predictions
    • Optimize existing machine learning algorithms to improve performance, reduce execution time and cost

    Expertise and Skills Needed:

    • Ensemble learning techniques
    • Deep learning
    • General machine learning for classification
    • Signal processing
    • Firmware development (optional)
    • Experience with IMUs, other wearable sensors

    For more info or to apply to this applied research position, please

    1. Check your eligibility and find more information about open projects.
    2. Interested students need to get the approval from their supervisor and send their CV along with a link to their supervisor’s university webpage by applying through the webform
    Program: