PrivatEye: Privacy-preserving eye movement data manipulation for virtual and augmented reality

• Applicant: Professorin Dr. Enkelejda Kasneci
• Subject Area: Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
• Term: since 2022
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 491966293

Eye-tracking is considered as a key emerging technology in virtual and augmented reality due to its potential to enable to improve user experience and save computational resources (e.g., through foveated rendering). Since eyes are not controlled fully consciously, one can also derive sensitive and personal information by analyzing eye movements. Consequently, biometric information included in the eye movement behavior must be protected from potential adversaries. However, so far, few studies have focused on privacy-preserving eye-tracking by proposing either differential privacy-based or practical solutions. These methods however suffer from several limitations. For instance, there is only one work that takes data correlations into account while providing differential privacy-based solutions. As standard differential privacy mechanisms are vulnerable to correlations in data, it is important to handle them properly. In addition, most of the proposed practical solutions do not offer formal privacy proofs nor have they been used in practice thus remaining proof-of-concepts. Furthermore, these solutions have not been evaluated across different visual stimuli and tasks. Hence, in this work, we aim to provide an overall framework by proposing differential privacy- and secure multi-party-based solutions and evaluating them in different application (i.e., education and driving). As the main outcome of this project, we envision privacy-preserving eye-tracking solutions that will be easy to deploy in real world VR/AR applications.

• DFG Programme: Research Grants