Aleksandr Beznosikov

Sign-SGD via Parameter-Free Optimization

Unlocking the Potential of Weighting Methods in Federated Learning Through Communication Compression

Convergence of Clipped-SGD for Convex $(L_0,L_1)$-Smooth Optimization with Heavy-Tailed Noise

Enhancing Stability of Physics-Informed Neural Network Training Through Saddle-Point Reformulation

Communication-Efficient Federated Learning with Adaptive Number of Participants

Leveraging Coordinate Momentum in SignSGD and Muon: Memory-Optimized Zero-Order LLM Fine-Tuning

Complexity-Separated Schemes for Addressing Structured Heterogeneity in Federated Learning

CompSGD: Robust Comparison-Based Approach for Zeroth-Order Optimization under $(L_0, L_1)$-Smoothness and Heavy-Tailed Noise

Aligning Distributionally Robust Optimization with Practical Deep Learning Needs

DIANA with Compression for Distributed Variational Inequalities: Eliminating the Need to Transmit Full Gradients

Preconditioned Norms: A Unified Framework for Steepest Descent, Quasi-Newton and Adaptive Methods

A Strategy-Agnostic Framework for Partial Participation in Federated Learning

Accelerated Methods with Complexity Separation Under Data Similarity for Federated Learning Problems

Johnson-Lindenstrauss Transforms in Distributed Optimization

Why SGD is not Brownian Motion: A New Perspective on Stochastic Dynamics

PRPO: Collaborative Online Policy Learning in Personalized RLHF

Unified Theory of Adaptive Variance Reduction

SimplexLoRA: Dynamic Rank Updating via Alternating Minimization with Simplex Projection

Power of Sign: High Probability Bounds Under $(L_0, L_1)$-smoothness and Heavy-Tailed Noise

Thinking like a CHEMIST: Combined Heterogeneous Embedding Model Integrating Structure and Tokens

Just a Simple Transformation is Enough for Data Protection in Split Learning

Dynamic Compression in Distributed Communications for Reduction of Transmission Losses

Diminishing Noise Maintains Differential Privacy and Enhances Convergence

A Robust Training Method for Federated Learning with Partial Participation

Gradient-Free Approaches is a Key to an Efficient Interaction with Markovian Stochasticity

Markovian Compression: Looking to the Past Helps Accelerate the Future

Decentralized Finite-Sum Optimization over Time-Varying Networks

Clipping Improves Adam and AdaGrad when the Noise Is Heavy-Tailed

FRUGAL: Memory-Efficient Optimization by Reducing State Overhead for Scalable Training

Exploring New Frontiers in Vertical Federated Learning: the Role of Saddle Point Reformulation

FRUGAL: Memory-Efficient Optimization by Reducing State Overhead for Scalable Training

Clipping Improves Adam-Norm and AdaGrad-Norm when the Noise Is Heavy-Tailed

Sign Operator for Coping with Heavy-Tailed Noise: High Probability Convergence Bounds with Extensions to Distributed Optimization and Comparison Oracle

Label Privacy in Split Learning for Large Models with Parameter-Efficient Training

Broadening Discovery through Structural Models: Multimodal Combination of Local and Structural Properties for Predicting Chemical Features.

Adaptive Regularized Newton Methods with Inexact Hessian