Train one machine-learning model across many parties who don't trust each other. A dishonest minority can't skew the result, and every party computes byte-for-byte identical output on any CPU or architecture (big-endian included) — so each round ships a receipt anyone can re-check offline and reproduce exactly. Rust; drop-in for Flower.
rust distributed-systems cross-platform fault-tolerance bft crdt reproducibility merkle-tree flower adversarial-machine-learning federated-learning determinism machine-learning-security verifiable-computation byzantine-fault-tolerance model-poisoning robust-aggregation byzantine-robustness gradient-aggregation krum
-
Updated
Sep 19, 2026 - Rust