VeilFL adds two-stage accountability to anonymous federated learning
Original title: VeilFL: Two-Stage Accountability for Anonymous Federated Learning
VeilFL detects duplicate submissions in PINE without persistent identifiers for normal participants using staged tracing.
VeilFL is a two-aggregator protocol that binds PINE messages with round-specific pseudonyms and tracing pairs for anonymous federated learning. After confirmed duplicates, a threshold tracing committee reveals a stable handle linking faults. Client identity is recoverable only after multiple linked faults and authorization. Prototype costs are low with batch verification.
- Detects duplicate submissions within a round
- Enables controlled two-stage opening
- Preserves PINE privacy guarantees
- Submission binding and duplicate detection
- Batch verification reduces cost to 1.86 ms per submission
From the original article
1 min read at arxiv.orgAnonymous credentials can hide which registered client participates in federated learning, but their unlinkability also makes repeated misconduct difficult to associate. We present VeilFL, a two-aggregator protocol that adds staged accountability to PINE (USENIX Security '24). Each submission binds two private PINE messages, a round-specific pseudonym, and a tracing pair to the same hidden credential secret.
By Varesh Mishra · 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.