40 citations · 57 across the 11 of their papers we have counts for
12 papers
Taming Nonconvex Stochastic Mirror Descent with General Bregman Divergence
Ilyas Fatkhullin, Niao He
This paper revisits the convergence of Stochastic Mirror Descent (SMD) in the contemporary nonconvex optimization setting. Existing results for batch-free nonconvex SMD restrict th…
Stochastic Optimization under Hidden Convexity
Ilyas Fatkhullin, Niao He, Yifan Hu
In this work, we consider constrained stochastic optimization problems under hidden convexity, i.e., those that admit a convex reformulation via non-linear (but invertible) map $c(…
Learning Zero-Sum Linear Quadratic Games with Improved Sample Complexity and Last-Iterate Convergence
Jiduan Wu, Anas Barakat, Ilyas Fatkhullin +1
Zero-sum Linear Quadratic (LQ) games are fundamental in optimal control and can be used (i)~as a dynamic game formulation for risk-sensitive or robust control and (ii)~as a benchma…
Reinforcement Learning with General Utilities: Simpler Variance Reduction and Large State-Action Space
Anas Barakat, Ilyas Fatkhullin, Niao He
We consider the reinforcement learning (RL) problem with general utilities which consists in maximizing a function of the state-action occupancy measure. Beyond the standard cumula…
Two Sides of One Coin: the Limits of Untuned SGD and the Power of Adaptive Methods
Junchi Yang, Xiang Li, Ilyas Fatkhullin +1
The classical analysis of Stochastic Gradient Descent (SGD) with polynomially decaying stepsize relies on well-tuned depending on problem parameters such as…
Momentum Provably Improves Error Feedback!
Ilyas Fatkhullin, Alexander Tyurin, Peter Richtárik
Due to the high communication overhead when training machine learning models in a distributed environment, modern algorithms invariably rely on lossy communication compression. How…
Stochastic Policy Gradient Methods: Improved Sample Complexity for Fisher-non-degenerate Policies
Ilyas Fatkhullin, Anas Barakat, Anastasia Kireeva +1
Recently, the impressive empirical success of policy gradient (PG) methods has catalyzed the development of their theoretical foundations. Despite the huge efforts directed at the…
Sharp Analysis of Stochastic Optimization under Global Kurdyka-Łojasiewicz Inequality
Ilyas Fatkhullin, Jalal Etesami, Niao He +1
We study the complexity of finding the global solution to stochastic nonconvex optimization when the objective function satisfies global Kurdyka-Lojasiewicz (KL) inequality and the…
3PC: Three Point Compressors for Communication-Efficient Distributed Training and a Better Theory for Lazy Aggregation
Peter Richtárik, Igor Sokolov, Ilyas Fatkhullin +3
We propose and study a new class of gradient communication mechanisms for communication-efficient training -- three point compressors (3PC) -- as well as efficient distributed nonc…
EF21 with Bells & Whistles: Six Algorithmic Extensions of Modern Error Feedback
Ilyas Fatkhullin, Igor Sokolov, Eduard Gorbunov +2
First proposed by Seide (2014) as a heuristic, error feedback (EF) is a very popular mechanism for enforcing convergence of distributed gradient-based optimization methods enhanced…
EF21: A New, Simpler, Theoretically Better, and Practically Faster Error Feedback
Peter Richtárik, Igor Sokolov, Ilyas Fatkhullin
Error feedback (EF), also known as error compensation, is an immensely popular convergence stabilization mechanism in the context of distributed training of supervised machine lear…
Optimizing Static Linear Feedback: Gradient Method
Ilyas Fatkhullin, Boris Polyak
The linear quadratic regulator is the fundamental problem of optimal control. Its state feedback version was set and solved in the early 1960s. However the static output feedback p…