Ilyas Fatkhullin
11 papers · 2021–2025 · 4 conferences · across top CS/AI conferences
Achievements
Jump to papers ↓+3 more ↓ Show less ↑
π Interdisciplinary Bridge π§ Keyword Pioneer π Renaissance Researcher (5) πΊοΈ Taxonomy Completionist (14) π Conference Polyglot (4)
π
Cross-Pollinator
(10)
π
Century Club
(11)
π₯
Unstoppable
(5)
Conferences
ICML (4)
NIPS (4)
AISTATS (2)
JMLR (1)
Top co-authors
Keywords
stochastic optimization
(5)
error feedback
(4)
nonconvex optimization
(4)
stochastic gradient descent
(3)
sample complexity
(3)
communication compression
(3)
variance reduction
(3)
distributed training
(2)
distributed learning
(2)
distributed optimization
(2)
policy gradient
(2)
convergence rate
(2)
gradient descent
(1)
stochastic approximation
(1)
linear function approximation
(1)
communication efficiency
(1)
stochastic mirror descent
(1)
convergence guarantee
(1)
gradient compression
(1)
gradient norm
(1)
Papers
EF21 with Bells & Whistles: Six Algorithmic Extensions of Modern Error Feedback
JMLR 2025
From Gradient Clipping to Normalization for Heavy Tailed SGD
AISTATS 2025
Safe-EF: Error Feedback for Non-smooth Constrained Optimization
ICML 2025
Taming Nonconvex Stochastic Mirror Descent with General Bregman Divergence
AISTATS 2024
Two Sides of One Coin: the Limits of Untuned SGD and the Power of Adaptive Methods
NIPS 2023
Reinforcement Learning with General Utilities: Simpler Variance Reduction and Large State-Action Space
ICML 2023
Stochastic Policy Gradient Methods: Improved Sample Complexity for Fisher-non-degenerate Policies
ICML 2023
Momentum Provably Improves Error Feedback!
NIPS 2023
3PC: Three Point Compressors for Communication-Efficient Distributed Training and a Better Theory for Lazy Aggregation
ICML 2022
Sharp Analysis of Stochastic Optimization under Global Kurdyka-Lojasiewicz Inequality
NIPS 2022
EF21: A New, Simpler, Theoretically Better, and Practically Faster Error Feedback
NIPS 2021