Privacy Preserving Eye-tracking Applications

The astounding advancements of technologies have not only brought benefits,  but have also introduced a range of potential risks. Among the most pressing issues is the critical challenge of privacy leakage, which requires urgent attention and systematic investigation. As a golden standard of biometrics, eye data is particularly vulnerable to privacy attacks. Unlike other biometrics, eye behavior is more pervasive, harder to control, and often related to subconsciousness. Eye movements also provide hints into users’ mental status and health condition, both physically and psychologically. Hence, in the context of XR, eye privacy is even more threatened, since eye-based interaction lays the foundation of HCI in virtual and augmented environments.

We aim at privacy-preserving applications of eye tracking using different techniques including differential privacy, federated learning, domain adaptation, randomized encryption and many more techniques. An emphasis of our work is the application in virtual environments. Part of this work will be integrated into our VR classroom project.

 

Further reading

• Efe Bozkir, David Geisler, and Enkelejda Kasneci. 2019. Person independent, privacy preserving, and real time assessment of cognitive load using eye tracking in a virtual reality setup. In 2019 IEEE Conference on Virtual Reality and 3D User Interfaces (VR). IEEE, 1834–1837.https://www.hci.uni-tuebingen.de/assets/pdf/publications/EDE032019.pdf
• Efe Bozkir, Onur Günlü, Wolfgang Fuhl, Rafael F. Schaefer, and Enkelejda Kasneci. 2021. Differential privacy for eye tracking with temporal correlations. Plos one 16, 8 (2021), e0255979. https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0255979
• Efe Bozkir, Ali Burak Ünal, Mete Akgün, Enkelejda Kasneci, and Nico Pfeifer. 2020. Privacy Preserving Gaze Estimation Using Synthetic Images via a Randomized Encoding Based Framework. In ACM Symposium on Eye Tracking Research and Applications (Stuttgart, Germany) (ETRA ’20 Short Papers). ACM, New York, NY, USA, Article 21, 5 pages. https://doi.org/10.1145/3379156.3391364
• Wolfgang Fuhl, Efe Bozkir, and Enkelejda Kasneci. 2021. Reinforcement learning for the privacy preservation and manipulation of eye tracking data. In Artificial Neural Networks and Machine Learning–ICANN 2021: 30th International Conference on Artificial Neural Networks, Bratislava, Slovakia, September 14–17, 2021, Proceedings, Part IV 30. Springer, 595–607.https://arxiv.org/pdf/2002.06806.pdf