Peng Zhao

Revisiting Matrix Sketching in Linear Bandits: Achieving Sublinear Regret via Dyadic Block Sketching

Gradient-Variation Online Adaptivity for Accelerated Optimization with Hölder Smoothness

Generalized Linear Bandits: Almost Optimal Regret with One-Pass Update

Optimistic Online-to-Batch Conversions for Accelerated Convergence and Universality

Parameter-free Algorithms for the Stochastically Extended Adversarial Model

Provably Efficient Online RLHF with One-Pass Reward Modeling

Heavy-Tailed Linear Bandits: Huber Regression with One-Pass Update

TreeLoRA: Efficient Continual Learning via Layer-Wise LoRAs Guided by a Hierarchical Gradient-Similarity Tree

Non-stationary Online Learning for Curved Losses: Improved Dynamic Regret via Mixability