Variational Autoencoder for Image-Based Augmentation of Eye-Tracking Data

Journal of Imaging

卷:7

Colm Loughnane¹  Federica Cilia²  Jean-Luc Guérin³  Romuald Carette³  Gilles Dequen³  Mahmoud Elbattah³ 

[1] Faculty of Science and Engineering, University of Limerick, V94 T9PX Limerick, Ireland;

[2] Laboratoire CRP-CPO, Université de Picardie Jules Verne, 80000 Amiens, France;

[3] Laboratoire Modélisation, Information, Systèmes (MIS), Université de Picardie Jules Verne, 80080 Amiens, France;

关键词: deep learning;    variational autoencoder;    data augmentation;    eye-tracking;   

DOI  :  10.3390/jimaging7050083

来源: DOAJ

【 摘 要 】

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.

【 授权许可】

Unknown