`

Timezone: »

 
EF21 with Bells & Whistles: Practical Algorithmic Extensions of Modern Error Feedback
Peter Richtarik · Igor Sokolov · Ilyas Fatkhullin · Eduard Gorbunov · Zhize Li
First proposed by Seide et al (2014) as a heuristic, error feedback (EF) is a very popular mechanism for enforcing convergence of distributed gradient-based optimization methods enhanced with communication compression strategies based on the application of contractive compression operators. However, existing theory of EF relies on very strong assumptions (e.g., bounded gradients), and provides pessimistic convergence rates (e.g., while the best known rate for EF in the smooth nonconvex regime, and when full gradients are compressed, is $O(1/T^{2/3})$, the rate of gradient descent in the same regime is $O(1/T)$). Recently, Richt\'{a}rik et al (2021) proposed a new error feedback mechanism, EF21, based on the construction of a Markov compressor induced by a contractive compressor. EF21 removes the aforementioned theoretical deficiencies of EF and at the same time works better in practice. In this work we propose six practical extensions of EF21, all supported by strong convergence theory: partial participation, stochastic approximation, variance reduction, proximal setting, momentum and bidirectional compression. Several of these techniques were never analyzed in conjunction with EF before, and in cases where they were (e.g., bidirectional compression), our rates are vastly superior.

Author Information

Peter Richtarik (KAUST)
Igor Sokolov (KAUST)
Ilyas Fatkhullin (TU Munich)
Eduard Gorbunov (Moscow Institute of Physics and Technology)
Zhize Li (King Abdullah University of Science and Technology (KAUST))

Zhize Li is a Research Scientist at the King Abdullah University of Science and Technology (KAUST) since September 2020. He obtained his PhD degree in Computer Science from Tsinghua University in 2019 (Advisor: Prof. Jian Li). He was a postdoc at KAUST (Hosted by Prof. Peter Richtárik), a visiting scholar at Duke University (Hosted by Prof. Rong Ge), and a visiting scholar at Georgia Institute of Technology (Hosted by Prof. Guanghui (George) Lan).

More from the Same Authors