Rondo and BumbleBee: Scalable Randomness Beacon and Secure Two‑Party Inference for Large Transformers

This article introduces two Ant Group papers selected for NDSS 2025—Rondo, a scalable, reconfiguration‑friendly distributed randomness beacon for blockchain‑type systems, and BumbleBee, a communication‑efficient two‑party inference framework that protects privacy for large Transformer models.

Secure AIlarge transformersprivacy-preserving inference

0 likes · 5 min read

Rondo and BumbleBee: Scalable Randomness Beacon and Secure Two‑Party Inference for Large Transformers

AntTech

Dec 6, 2024 · Artificial Intelligence

Nimbus: Secure and Efficient Two‑Party Inference for Transformers

The paper introduces Nimbus, a two‑party privacy‑preserving inference framework for Transformer models that leverages a client‑side outer‑product linear‑layer protocol and distribution‑aware polynomial approximations for non‑linear layers, achieving up to five‑fold speedups with negligible accuracy loss.

Homomorphic EncryptionPerformance OptimizationTransformer

0 likes · 15 min read

Nimbus: Secure and Efficient Two‑Party Inference for Transformers