Variational Autoencoder for Image-Based Augmentation of Eye-Tracking Data
Over the past decade, deep learning has achieved unprecedented successes in a diversity of application domains, given large-scale datasets. However, particular domains, such as healthcare, inherently suffer from data paucity and imbalance. Moreover, datasets could be largely inaccessible due to priv...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-05-01
|
| Series: | Journal of Imaging |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2313-433X/7/5/83 |
| _version_ | 1797535350189457408 |
|---|---|
| author | Mahmoud Elbattah Colm Loughnane Jean-Luc Guérin Romuald Carette Federica Cilia Gilles Dequen |
| author_facet | Mahmoud Elbattah Colm Loughnane Jean-Luc Guérin Romuald Carette Federica Cilia Gilles Dequen |
| author_sort | Mahmoud Elbattah |
| collection | DOAJ |
| description | Over the past decade, deep learning has achieved unprecedented successes in a diversity of application domains, given large-scale datasets. However, particular domains, such as healthcare, inherently suffer from data paucity and imbalance. Moreover, datasets could be largely inaccessible due to privacy concerns, or lack of data-sharing incentives. Such challenges have attached significance to the application of generative modeling and data augmentation in that domain. In this context, this study explores a machine learning-based approach for generating synthetic eye-tracking data. We explore a novel application of variational autoencoders (VAEs) in this regard. More specifically, a VAE model is trained to generate an image-based representation of the eye-tracking output, so-called scanpaths. Overall, our results validate that the VAE model could generate a plausible output from a limited dataset. Finally, it is empirically demonstrated that such approach could be employed as a mechanism for data augmentation to improve the performance in classification tasks. |
| first_indexed | 2024-03-10T11:43:13Z |
| format | Article |
| id | doaj.art-3f6851a488504a5fb9ef0d7de2f45870 |
| institution | Directory Open Access Journal |
| issn | 2313-433X |
| language | English |
| last_indexed | 2024-03-10T11:43:13Z |
| publishDate | 2021-05-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Journal of Imaging |
| spelling | doaj.art-3f6851a488504a5fb9ef0d7de2f458702023-11-21T18:15:27ZengMDPI AGJournal of Imaging2313-433X2021-05-01758310.3390/jimaging7050083Variational Autoencoder for Image-Based Augmentation of Eye-Tracking DataMahmoud Elbattah0Colm Loughnane1Jean-Luc Guérin2Romuald Carette3Federica Cilia4Gilles Dequen5Laboratoire Modélisation, Information, Systèmes (MIS), Université de Picardie Jules Verne, 80080 Amiens, FranceFaculty of Science and Engineering, University of Limerick, V94 T9PX Limerick, IrelandLaboratoire Modélisation, Information, Systèmes (MIS), Université de Picardie Jules Verne, 80080 Amiens, FranceLaboratoire Modélisation, Information, Systèmes (MIS), Université de Picardie Jules Verne, 80080 Amiens, FranceLaboratoire CRP-CPO, Université de Picardie Jules Verne, 80000 Amiens, FranceLaboratoire Modélisation, Information, Systèmes (MIS), Université de Picardie Jules Verne, 80080 Amiens, FranceOver the past decade, deep learning has achieved unprecedented successes in a diversity of application domains, given large-scale datasets. However, particular domains, such as healthcare, inherently suffer from data paucity and imbalance. Moreover, datasets could be largely inaccessible due to privacy concerns, or lack of data-sharing incentives. Such challenges have attached significance to the application of generative modeling and data augmentation in that domain. In this context, this study explores a machine learning-based approach for generating synthetic eye-tracking data. We explore a novel application of variational autoencoders (VAEs) in this regard. More specifically, a VAE model is trained to generate an image-based representation of the eye-tracking output, so-called scanpaths. Overall, our results validate that the VAE model could generate a plausible output from a limited dataset. Finally, it is empirically demonstrated that such approach could be employed as a mechanism for data augmentation to improve the performance in classification tasks.https://www.mdpi.com/2313-433X/7/5/83deep learningvariational autoencoderdata augmentationeye-tracking |
| spellingShingle | Mahmoud Elbattah Colm Loughnane Jean-Luc Guérin Romuald Carette Federica Cilia Gilles Dequen Variational Autoencoder for Image-Based Augmentation of Eye-Tracking Data Journal of Imaging deep learning variational autoencoder data augmentation eye-tracking |
| title | Variational Autoencoder for Image-Based Augmentation of Eye-Tracking Data |
| title_full | Variational Autoencoder for Image-Based Augmentation of Eye-Tracking Data |
| title_fullStr | Variational Autoencoder for Image-Based Augmentation of Eye-Tracking Data |
| title_full_unstemmed | Variational Autoencoder for Image-Based Augmentation of Eye-Tracking Data |
| title_short | Variational Autoencoder for Image-Based Augmentation of Eye-Tracking Data |
| title_sort | variational autoencoder for image based augmentation of eye tracking data |
| topic | deep learning variational autoencoder data augmentation eye-tracking |
| url | https://www.mdpi.com/2313-433X/7/5/83 |
| work_keys_str_mv | AT mahmoudelbattah variationalautoencoderforimagebasedaugmentationofeyetrackingdata AT colmloughnane variationalautoencoderforimagebasedaugmentationofeyetrackingdata AT jeanlucguerin variationalautoencoderforimagebasedaugmentationofeyetrackingdata AT romualdcarette variationalautoencoderforimagebasedaugmentationofeyetrackingdata AT federicacilia variationalautoencoderforimagebasedaugmentationofeyetrackingdata AT gillesdequen variationalautoencoderforimagebasedaugmentationofeyetrackingdata |