A trimmed product-of-experts aggregation plus a local/global fusion rule gives online Gaussian process regression with error bounds under fewer than 25% Byzantine agents.
Title resolution pending
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
-
Byzantine-resilient federated online learning for Gaussian process regression
A trimmed product-of-experts aggregation plus a local/global fusion rule gives online Gaussian process regression with error bounds under fewer than 25% Byzantine agents.