Introduces adversarially replicable online learning, with sublinear-regret algorithms for linear optimization and experts, plus regret lower bounds.
Large deviations for martingales
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Replicable Online Learning
Introduces adversarially replicable online learning, with sublinear-regret algorithms for linear optimization and experts, plus regret lower bounds.