A-GE estimates age from device motion without identity data
Original title: A-GE: Privacy-Preserving Age Assurance via Fine-Grained Motion
Across more than 600 participants, a 16-year threshold reached up to 93.8% accuracy.
A-GE derives age-related evidence from fine-grained motion captured by commodity inertial sensors, without identity documents or an identity provider. Thresholds are configurable from 14 to 21 years, including web sensing near 60 Hz. Evaluation used more than 600 participants in disjoint development and test cohorts. Client-side deployment is shown on Android, iOS, and the web, and the authors also examine unintended attribute inference.
Unconfirmed: The claim that motion reflects neuromuscular and physiological processes is stated as a hypothesis.
- At a 16-year threshold, accuracy reached up to 93.8% with a comparable F1 score.
- Model development and final evaluation used separate participant-disjoint cohorts.
- The system runs at available sensor rates, including about 60 Hz on the web.
- Authors examine residual privacy risks from commodity motion features.
- Products
- A-GE
From the original article
1 min read at arxiv.orgAge assurance faces a fundamental privacy dilemma: meaningful assurance typically requires revealing identity or other sensitive information, whereas privacy-preserving alternatives often provide little assurance. Existing approaches either disclose sensitive information directly or shift the disclosure and trust requirement to an identity provider, while self-declared age provides little meaningful assurance.
By Mohammed Gharib · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.