Score: 0

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

Published: January 13, 2026 | arXiv ID: 2601.08614v1

By: Dmitry Bylinkin , Sergey Skorik , Dmitriy Bystrov and more

Potential Business Impact:

Makes computers learn together without sharing private data.

Business Areas:
A/B Testing Data and Analytics

Heterogeneity within data distribution poses a challenge in many modern federated learning tasks. We formalize it as an optimization problem involving a computationally heavy composite under data similarity. By employing different sets of assumptions, we present several approaches to develop communication-efficient methods. An optimal algorithm is proposed for the convex case. The constructed theory is validated through a series of experiments across various problems.

Page Count
30 pages

Category
Mathematics:
Optimization and Control