Zhao-Jun Peng

This work introduces Decision-Relevant Fresh Comparison (DRFC), which uses new, balanced samples from all agents to compare arms at the network level and switches only when fresh global evidence indicates that the common best arm has changed, and proves a high-probability dynamic regret bound with no adaptation term depending on $\Stloc$.

Achieving sample complexity bounds that are independent of mixing time could revolutionize how we approach stochastic optimization in nonconvex settings.