Introduces a novel variance-reduced gradient estimator using Bernoulli distribution for single-direction renovation or full correction, integrated with gradient tracking, achieving oracle complexities O(d/ε) for smooth nonconvex and O(dκ ln(1/ε)) for gradient-dominated functions.
Horn and Charles R
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
verdicts
UNVERDICTED 2representative citing papers
Variance-aware neural dueling bandit algorithms achieve sublinear regret of order O(d sqrt(sum sigma_t^2) + sqrt(d T)) for wide networks on nonlinear utilities.
citing papers explorer
-
Variance-Reduced Gradient Estimator for Nonconvex Zeroth-Order Distributed Optimization
Introduces a novel variance-reduced gradient estimator using Bernoulli distribution for single-direction renovation or full correction, integrated with gradient tracking, achieving oracle complexities O(d/ε) for smooth nonconvex and O(dκ ln(1/ε)) for gradient-dominated functions.
-
Neural Variance-aware Dueling Bandits with Deep Representation and Shallow Exploration
Variance-aware neural dueling bandit algorithms achieve sublinear regret of order O(d sqrt(sum sigma_t^2) + sqrt(d T)) for wide networks on nonlinear utilities.