CAFCOR combines pairwise shared-randomness correlated noise with a covariance-agnostic robust filter to achieve central-DP-like utility under the SecLDP trust model with Byzantine workers and an untrusted server.
Robust distributed learning: Tight error bounds and breakdown point under data heterogeneity
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Towards Trustworthy Federated Learning with Untrusted Participants
CAFCOR combines pairwise shared-randomness correlated noise with a covariance-agnostic robust filter to achieve central-DP-like utility under the SecLDP trust model with Byzantine workers and an untrusted server.