Yuriy Nesterov
PulseAugur coverage of Yuriy Nesterov — every cluster mentioning Yuriy Nesterov across labs, papers, and developer communities, ranked by signal.
3 day(s) with sentiment data
-
New theoretical bounds for high-order convex optimization established
Researchers have established tight lower bounds for optimizing functions that exhibit both high-order Hölder smoothness and uniform convexity, specifically when these properties match. The study, which references the wo…
-
Nesterov trajectories are locally minimizing only for p=4, new paper finds
A new research paper explores the conditions under which Nesterov's optimization method produces trajectories that are eventually locally minimizing. The study establishes a connection between this property and the nonn…
-
New attention mechanism inspired by Hamiltonian dynamics detailed in arXiv paper
Researchers have developed SympFormer, a novel attention block inspired by damped Hamiltonian dynamics on probability measure spaces. This approach extends the understanding of attention mechanisms in large language mod…
-
AF-Muon optimizer eliminates AdamW for tied-embedding models, saving memory
Researchers have developed AF-Muon, an optimizer that removes the need for AdamW in tied-embedding models, potentially saving around 20% of optimizer-state memory. This new approach maintains Muon's spectral-norm steepe…
-
New paper details Geometric Moment Contraction for Stochastic Nesterov Acceleration
A new paper published on arXiv details a method called Geometric Moment Contraction (GMC) for improving Stochastic Nesterov Acceleration. The research introduces a direct criterion for synchronous L^p contraction under …
-
New distributed algorithms accelerate monotone inclusion solutions
Researchers have developed two novel distributed algorithms, ND-DFFP and NI-DFFP, designed to efficiently solve complex monotone inclusion problems over networks. These algorithms integrate Nesterov-type acceleration wi…
-
New accelerated algorithms improve convergence for generalized equations
Researchers have developed a new algorithmic framework that combines Nesterov's acceleration and variance-reduction techniques to solve a class of generalized equations. This method is designed for data-driven applicati…
-
Gradient Descent Dynamics Explored in New Optimization Research
Two new arXiv papers delve into the complexities of gradient descent algorithms. The first paper by Si Yi Meng examines gradient descent dynamics on logistic regression with non-separable data and large step sizes, reve…
-
New research details momentum's role in large-batch AI training
A new research paper explores how momentum impacts large-batch training in machine learning, using power-law kernel regression as a model. The study defines a critical learning rate to characterize risk stability and de…
-
Federated learning research tackles Byzantine attacks with new algorithms and placement strategies
Researchers are exploring enhanced security and efficiency in federated learning, particularly against Byzantine attacks where malicious participants can corrupt the training process. One study introduces a Nesterov-acc…
-
New discrete state sampling algorithms leverage Nesterov's acceleration
Researchers have developed a new class of algorithms for sampling discrete states, building upon Nesterov's accelerated gradient method. This approach extends the traditional Metropolis-Hastings algorithm by interpretin…
-
New Nesterov acceleration methods developed for probability measures
Researchers have developed new accelerated optimization methods for probability measures, drawing inspiration from Nesterov's accelerated gradient method in Euclidean space. These methods, including Heavy-ball and Neste…
-
New OptMuon method enhances stochastic optimization with adaptive momentum
Researchers have introduced OptMuon, a novel adaptive momentum orthogonalization method for stochastic nonconvex optimization that calibrates update magnitudes from observed trajectories. This approach combines Muon-sty…
-
Researchers develop online Newton method with accelerated sketching for efficient inference
A new paper introduces an online Newton method that uses Nesterov's accelerated sketching to approximate Newton directions. This approach aims to provide robust uncertainty quantification for streaming data while mainta…