Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Internship or Master Thesis: Machine learning for the classification of shoulder loading activities in real-life of wheelchair users
Around 40% of persons with SCI in Switzerland report shoulder pain, and even higher proportions suffer from shoulder pathologies, limiting in their mobility and participation. Shoulder overuse is seen as a major contributor to such shoulder complaints, however, little is known about the requirements of daily life with respect to shoulder load, and risk factors for overuse. Currently, a proof of concept methodology has been developed for the classification of wheelchair-related activities, based on wearable sensor data (www.mdpi.com/1424-8220/22/19/7404).
However, further research is required to achieve a usable algorithm for real-life environments. We will focus on 1-class classifiers and unsupervised clustering for exploring higher accuracies in unknown situations.
Using Several datasets collected over the past years for validation of novel classification methods that could be exported to real-life.
A new methodology should be developed and new testing data collected, adapted or developed.
You will also have the opportunity to participate/support ongoing measurement at the motion laboratory, and in real-life settings.
The successful student candidate will be affiliated to SPF’s Shoulder Health and Mobility Group, for a master thesis of 6 months (or meeting your institutional requirements), starting at earliest convenience, or upon agreement.
Using Several datasets collected over the past years for validation of novel classification methods that could be exported to real-life. A new methodology should be developed and new testing data collected, adapted or developed.
You will also have the opportunity to participate/support ongoing measurement at the motion laboratory, and in real-life settings.
The successful student candidate will be affiliated to SPF’s Shoulder Health and Mobility Group, for a master thesis of 6 months (or meeting your institutional requirements), starting at earliest convenience, or upon agreement.
The goal is to further develop an approach to activity classification that could be ported to real-life data.
This means the algorithm should be able to classify the list of relevant activities we are interested in and identify any other activity as "Not of Interest". The most likely solution here is the combination of unsupervised clustering of continuous sensor data and supervised classification of a subset thereof.
But it is up to the master student to find and develop a working solution!!!
The goal is to further develop an approach to activity classification that could be ported to real-life data. This means the algorithm should be able to classify the list of relevant activities we are interested in and identify any other activity as "Not of Interest". The most likely solution here is the combination of unsupervised clustering of continuous sensor data and supervised classification of a subset thereof. But it is up to the master student to find and develop a working solution!!!
For further information please contact Dr. Diego Paez (Head of SCAI Lab) diego.paez@resc.ethz.ch
For further information please contact Dr. Diego Paez (Head of SCAI Lab) diego.paez@resc.ethz.ch