A flexible-sampling GDRO algorithm achieves O(1/t sqrt(sum_j m/r_j log m)) high-probability optimization error, generalizing prior r=1 and r=m guarantees.
Regret analysis of stochastic and nonstochastic multi-armed bandit problems
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
-
Group Distributionally Robust Optimization with Flexible Sample Queries
A flexible-sampling GDRO algorithm achieves O(1/t sqrt(sum_j m/r_j log m)) high-probability optimization error, generalizing prior r=1 and r=m guarantees.