Difference between revisions of "Subject specific embeddings for transfer learning in brain-computer interfaces"
From iis-projects
Line 1: | Line 1: | ||
− | [[Category:Digital | + | [[Category:Digital]] |
[[File:Emotiv-epoc-14-channel-mobile-eeg.jpg|thumb|300px]] | [[File:Emotiv-epoc-14-channel-mobile-eeg.jpg|thumb|300px]] | ||
==Description== | ==Description== |
Latest revision as of 11:12, 21 July 2020
Contents
Description
A brain–computer interface is a device that enables communication and control without movement. The device aims to recognize the human’s intentions from spatiotemporal neural activity typically recorded by a large set of electroencephalogram (EEG) electrodes. What makes it particularly challenging, however, is its susceptibility to errors in the recognition of human intentions. The underlying reason is the low signal to noise ratio due to high inter-subject variance, which makes it difficult to build one universal model for all subjects.
This project aims to overcome this issue by adding trainable subject specific embeddings to the model. You can start from an existing CNN implementation [1], and extend it with additional embeddings.
Status: Available
- Looking for 2 students for a semester project or 1 student for a master thesis.
- Supervision: Michael Hersche, Lukas Cavigelli
Prerequisites
- Machine Learning
- Linear Algebra
- Python Programming
Character
- 20% Theory
- 80% Programming
Professor
Literature
- [1] Schirrmeister et. al., Deep learning with convolutional neural networks for EEG decoding and visualization