2 papers
cs.LG2026
The Cost of an Unknown Horizon for the Last Iterate of Clipped SGD under Heavy-Tailed Noise
Yuxing Peng, Zhiqing Tang, Weijia Jia
We study the last iterate of clipped SGD, and of every rule that queries its current iterate and steps along the resulting stochastic gradient, when training is stopped at a data-d…
cs.DS2026
Dense Weak Hiding: Closing Complexity Gaps in Nonconvex and PL Finite-Sum Optimization under Individual Smoothness
Yuxing Peng, Zhiqing Tang, Weijia Jia
Under individual smoothness, known algorithms for nonconvex finite-sum optimization use incremental first-order oracle (IFO) calls, while exis…