Hidden In Plain Gaze: Gaze Representations as Privacy Controls for Utility and Re-identification Risk in XR

SummaryAI generated

On September 7, 2026, arXiv cs.LG published the paper “Hidden In Plain Gaze”, proposing to use gaze representations as a privacy control mechanism in XR applications. The study analyzed the balance between improving system efficiency and posing risks of reidentification with gaze data. Its goal is to enhance user privacy protection by optimizing gaze representations while maintaining the practicality and security of XR interactions.

Related eventsRELATED EVENTS

2026-09-09 05:29OpenAI claims huge maths breakthrough on a famed ‘Millennium Problem’
2026-09-09 06:11OpenAI claims to have solved the Navier-Stokes equation problem, but its originality is questioned by the academic community
2026-09-08 22:00Google DeepMind Maps 9 Billion Possible DNA Variants

Key entitiesKEY ENTITIES

HoloAssist

SignalsSIGNALS

Keyword heat

HoloAssist1

All reports (1)SOURCES

↗

A arXiv cs.LG en 2026-09-07 12:00

研究团队利用 egocentric HoloAssist 数据集，对比了 raw gaze、spatial attention heatmaps 和 engineered eye-movement features 三种 gaze representations。在保持模型容量的前提下，engineered features 将 re-identification 风险降低了约一个数量级（降至 206 名身份的四倍），同时保留了约 85% 的 action recognition 准确率。结果显示，仅通过 abstraction 无法完全消除 identity leakage，但 engineered features 提供了可解释且可审计的结构，为设计师提供了透明隐私控制手段，以补充 differentia…