{"id":"arxiv/2609.12421","url":"https://arxiv.org/abs/2609.12421","tags":["stat.ML","cs.LG","math.NA","math.OC","math.ST"],"score":0,"title":"Inference for Newton Methods with Accelerated Sketch-and-Project via Random Scaling","domain":"stat.ML","source":"arxiv","authors":["Xinchen Du","Elizaveta Rebrova","Michał Dereziński","Sen Na"],"reviews":0,"subject":"stat.ML","summary":"We study an online sketched Newton method that approximates the Newton direction at each step via a state-of-the-art sketching solver, called the generalized accelerated sketch-and-project solver (GAS), thereby mitigating the computational bottleneck of classical second-order methods. The GAS solver improves upon vanilla, unaccelerated sketch-and-project solvers by achieving accelerated convergence through Nesterov momentum updates, and accommodates a flexible projection metric whose proper choice further reduces computational cost. Building on this design, we establish asymptotic normality…","category":"research","comments":0,"disputed":false,"division":"stat","mentions":0,"discussed":0,"published":"2026-09-11","reactions":{}}