Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

FEW-SHOT CLASSIFICATION OF EEG WITH QUASI-INDUCTIVE TRANSFER LEARNING

Duedahl, Mathias LU (2022) In Bachelor's Theses in Mathematicas Sciences MASK11 20222
Mathematical Statistics
Abstract
Brain-computer interfaces (BCIs) are devices that enable people with disabilities to use their thoughts to control external devices and restore or improve
their bodily functions. One important aspect of BCIs is the classification of
electroencephalography (EEG) signals, which measure brain activity and can
be difficult to interpret. To address this challenge, we use time-frequency
transformations to convert EEG signals into images and employ pre-trained
deep convolutional neural networks (CNNs) to classify the images based on
whether the subject heard a sound from the left or the right ear. We investigate whether transfer learning, a technique that involves using a pre-trained
model on a related task as the starting point for... (More)
Brain-computer interfaces (BCIs) are devices that enable people with disabilities to use their thoughts to control external devices and restore or improve
their bodily functions. One important aspect of BCIs is the classification of
electroencephalography (EEG) signals, which measure brain activity and can
be difficult to interpret. To address this challenge, we use time-frequency
transformations to convert EEG signals into images and employ pre-trained
deep convolutional neural networks (CNNs) to classify the images based on
whether the subject heard a sound from the left or the right ear. We investigate whether transfer learning, a technique that involves using a pre-trained
model on a related task as the starting point for training on a new task, is
effective even when the source and target domains are very different. The
best classification result achieved was 61.7%, using EfficientNet V2 tuned on
5 different test-subjects, and tuned on the target subject. (Less)
Popular Abstract
Brain-computer interfaces (BCIs) are devices that enable people with disabilities to use their thoughts to control external devices and restore or improve
their bodily functions. One important aspect of BCIs is the classification of
electroencephalography (EEG) signals, which measure brain activity and can
be difficult to interpret. To address this challenge, we use time-frequency
transformations to convert EEG signals into images and employ pre-trained
deep convolutional neural networks (CNNs), networks specialized in image classification, to classify the images based on
whether the subject heard a sound from the left or the right ear. We investigate whether transfer learning, a technique that involves using a pre-trained
model... (More)
Brain-computer interfaces (BCIs) are devices that enable people with disabilities to use their thoughts to control external devices and restore or improve
their bodily functions. One important aspect of BCIs is the classification of
electroencephalography (EEG) signals, which measure brain activity and can
be difficult to interpret. To address this challenge, we use time-frequency
transformations to convert EEG signals into images and employ pre-trained
deep convolutional neural networks (CNNs), networks specialized in image classification, to classify the images based on
whether the subject heard a sound from the left or the right ear. We investigate whether transfer learning, a technique that involves using a pre-trained
model with knowledge of one task to more easily acquire knowledge of a related task, is effective even when the source and target domains are very different. The
best classification result achieved was 61.7%, using EfficientNet V2 tuned on
5 different test-subjects, and tuned on the target subject. (Less)
Please use this url to cite or link to this publication:
author
Duedahl, Mathias LU
supervisor
organization
course
MASK11 20222
year
type
M2 - Bachelor Degree
subject
keywords
Few-Shot Learning, Transfer Learning, EEG Classification, Deep Convolutional Neural Network, DNN, CNN, EEG, FSL, TL
publication/series
Bachelor's Theses in Mathematicas Sciences
report number
LUNFMS-4068-2022
ISSN
1654-6229
other publication id
2022:K22
language
English
id
9104237
date added to LUP
2023-01-17 14:21:11
date last changed
2023-01-17 14:21:11
@misc{9104237,
  abstract     = {{Brain-computer interfaces (BCIs) are devices that enable people with disabilities to use their thoughts to control external devices and restore or improve
their bodily functions. One important aspect of BCIs is the classification of
electroencephalography (EEG) signals, which measure brain activity and can
be difficult to interpret. To address this challenge, we use time-frequency
transformations to convert EEG signals into images and employ pre-trained
deep convolutional neural networks (CNNs) to classify the images based on
whether the subject heard a sound from the left or the right ear. We investigate whether transfer learning, a technique that involves using a pre-trained
model on a related task as the starting point for training on a new task, is
effective even when the source and target domains are very different. The
best classification result achieved was 61.7%, using EfficientNet V2 tuned on
5 different test-subjects, and tuned on the target subject.}},
  author       = {{Duedahl, Mathias}},
  issn         = {{1654-6229}},
  language     = {{eng}},
  note         = {{Student Paper}},
  series       = {{Bachelor's Theses in Mathematicas Sciences}},
  title        = {{FEW-SHOT CLASSIFICATION OF EEG WITH QUASI-INDUCTIVE TRANSFER LEARNING}},
  year         = {{2022}},
}