Convex

TL;DR: A reformulation of gradient descent based on a Sobolev-Slobodeckij inner product enables us to make rapid progress toward local minima—independent of curve resolution, and a hierarchical multigrid scheme that significantly reduces the per-step cost of optimization.

TL;DR: By adopting the implicit stochastic gradient descent algorithm to optimize the objective function for prediction, multiple matrices from different domains are consolidated inside the cross-domain recommendation system (CDRS), and a conceptual framework for CDRS is designed.

TL;DR: The AutoFi is proposed, an automatic WiFi sensing model based on a novel geometric self-supervised learning algorithm that transfers knowledge from randomly collected CSI samples into human gait recognition and achieves state-of-the-art performance.

TL;DR: PI++ builds on prediction-powered inference (PPI), which targets the same problem setting, improving its computational and statistical efficiency and easy-to-compute confidence sets.

TL;DR: This paper proposes a stochastic nested variance reduction algorithm for nonconvex optimization, achieving improved convergence rates and gradient complexity compared to existing methods, including SVRG and SCSG, with thorough experimental results supporting the theory.

TL;DR: This paper proposes SARAH, a novel stochastic recursive gradient algorithm for finite-sum minimization problems, offering a simple recursive framework and linear convergence rate under strong convexity, outperforming existing methods like SVRG and SAG/SAGA in numerical experiments.

TL;DR: This paper introduces a variance reduction method for non-convex optimization, achieving an O(1/ε) convergence rate for smooth objectives, outperforming full gradient descent by Ω(n^1/3), and demonstrating effectiveness on empirical risk minimizations and neural net training.