KM
Konstantin Mishchenko
math.OCcs.LGstat.MLcs.AIcs.NAmath.NAcs.DCcs.CL
On Valency
published · living versionsW_5txyfx4z·v1 · currentpublished
Preprints & journals
38 papers in the corpus · 2018–2025Smoothing DiLoCo with Primal Averaging for Faster Training of LLMs2512.17131v3 · Aaron Defazio, Konstantin Mishchenko, Parameswaran Raman et al.2025 · 0 citationsarXiv
Analysis of an Idealized Stochastic Polyak Method and its Application to Black-Box Model Distillation2504.01898v1 · Robert M. Gower, Guillaume Garrigos, Nicolas Loizou et al.2025 · 0 citationsarXiv
Hardware-Aware Parallel Prompt Decoding for Memory-Efficient Acceleration of LLM Inference2405.18628v3 · Hao Mark Chen, Wayne Luk, Ka Fai Cedric Yiu et al.2024 · 8 citationsarXiv
The Road Less Scheduled2405.15682v4 · Aaron Defazio, Xingyu Alice Yang, Harsh Mehta et al.2024 · 27 citationsarXiv
Optimal Linear Decay Learning Rate Schedules and Further Refinements2310.07831v2 · Aaron Defazio, Ashok Cutkosky, Harsh Mehta et al.2023 · 5 citationsarXiv
Adaptive Proximal Gradient Method for Convex Optimization2308.02261v2 · Yura Malitsky, Konstantin Mishchenko2023 · 16 citationsarXiv
Prodigy: An Expeditiously Adaptive Parameter-Free Learner2306.06101v4 · Konstantin Mishchenko, Aaron Defazio2023 · 3 citationsarXiv
Partially Personalized Federated Learning: Breaking the Curse of Data Heterogeneity2305.18285v1 · Konstantin Mishchenko, Rustem Islamov, Eduard Gorbunov et al.2023 · 3 citationsarXiv
DoWG Unleashed: An Efficient Universal Parameter-Free Gradient Descent Method2305.16284v4 · Ahmed Khaled, Konstantin Mishchenko, Chi Jin2023 · 4 citationsarXiv
Two Losses Are Better Than One: Faster Optimization Using a Cheaper Proxy2302.03542v3 · Blake Woodworth (SIERRA), Konstantin Mishchenko, Francis Bach (SIERRA et al.2023 · 0 citationsarXiv
Learning-Rate-Free Learning by D-Adaptation2301.07733v5 · Aaron Defazio, Konstantin Mishchenko2023 · 8 citationsarXiv
Convergence of First-Order Algorithms for Meta-Learning with Moreau Envelopes2301.06806v1 · Konstantin Mishchenko, Slavom'ir Hanzely, Peter Richt'arik2023 · 0 citationsarXiv
Super-Universal Regularized Newton Method2208.05888v1 · Nikita Doikov, Konstantin Mishchenko, Yurii Nesterov2022 · 16 citationsarXiv
Adaptive Learning Rates for Faster Stochastic Gradient Methods2208.05287v1 · Samuel Horv'ath, Konstantin Mishchenko, Peter Richt'arik2022 · 3 citationsarXiv
Asynchronous SGD Beats Minibatch SGD Under Arbitrary Delays2206.07638v2 · Konstantin Mishchenko, Francis Bach, Mathieu Even et al.2022 · 23 citationsarXiv
ProxSkip: Yes! Local Gradient Steps Provably Lead to Communication Acceleration! Finally!2202.09357v2 · Konstantin Mishchenko, Grigory Malinovsky, Sebastian Stich et al.2022 · 34 citationsarXiv
Server-Side Stepsizes and Sampling Without Replacement Provably Help in Federated Optimization2201.11066v1 · Grigory Malinovsky, Konstantin Mishchenko, Peter Richt'arik2022 · 5 citationsarXiv
Regularized Newton Method with Global Convergence2112.02089v3 · Konstantin Mishchenko2021 · 4 citationsarXiv
On Seven Fundamental Optimization Challenges in Machine Learning2110.12281v1 · Konstantin Mishchenko2021 · 1 citationarXiv
IntSGD: Adaptive Floatless Compression of Stochastic Gradients2102.08374v2 · Konstantin Mishchenko, Bokun Wang, Dmitry Kovalev et al.2021 · 10 citationsInternational Conference on Learning Representations (2022)
Proximal and Federated Random Reshuffling2102.06704v1 · Konstantin Mishchenko, Ahmed Khaled, Peter Richt'arik2021 · 7 citationsarXiv
Random Reshuffling: Simple Analysis with Vast Improvements2006.05988v3 · Konstantin Mishchenko, Ahmed Khaled, Peter Richt'arik2020 · 32 citationsarXiv
Dualize, Split, Randomize: Toward Fast Nonsmooth Optimization Algorithms2004.02635v4 · Adil Salim, Laurent Condat, Konstantin Mishchenko et al.2020 · 19 citationsarXiv
Stochastic Newton and Cubic Newton Methods with Simple Local Linear-Quadratic Rates1912.01597v1 · Dmitry Kovalev, Konstantin Mishchenko, Peter Richt'arik2019 · 15 citationsarXiv
Adaptive Gradient Descent without Descent1910.09529v2 · Yura Malitsky, Konstantin Mishchenko2019 · 40 citationsProceedings of the 37-th International Conference on Machine Learning, Online, PMLR 119, 2020
Sinkhorn Algorithm as a Special Case of Stochastic Mirror Descent1909.06918v1 · Konstantin Mishchenko2019 · 3 citationsarXiv
First Analysis of Local GD on Heterogeneous Data1909.04715v2 · Ahmed Khaled, Konstantin Mishchenko, Peter Richt'arik2019 · 103 citationsarXiv
Tighter Theory for Local SGD on Identical and Heterogeneous Data1909.04746v4 · Ahmed Khaled, Konstantin Mishchenko, Peter Richt'arik2019 · 162 citationsarXiv
A Self-supervised Approach to Hierarchical Forecasting with Applications to Groupwise Synthetic Controls1906.10586v1 · Konstantin Mishchenko, Mallory Montgomery, Federico Vaggi2019 · 10 citationsarXiv
MISO is Making a Comeback With Better Proofs and Rates1906.01474v1 · Xun Qian, Alibek Sailanbayev, Konstantin Mishchenko et al.2019 · 5 citationsarXiv
Revisiting Stochastic Extragradient1905.11373v2 · Konstantin Mishchenko, Dmitry Kovalev, Egor Shulgin et al.2019 · 13 citationsProceedings of the Twenty Third International Conference on Artificial Intelligence and Statistics, PMLR 108:4573-4582, 2020
A Stochastic Decoupling Method for Minimizing the Sum of Smooth and Non-Smooth Functions1905.11535v2 · Konstantin Mishchenko, Peter Richt'arik2019 · 7 citationsarXiv
Stochastic Distributed Learning with Gradient Quantization and Variance Reduction1904.05115v1 · Samuel Horv'ath, Dmitry Kovalev, Konstantin Mishchenko et al.2019 · 75 citationsarXiv
99% of Distributed Optimization is a Waste of Time: The Issue and How to Fix it1901.09437v2 · Konstantin Mishchenko, Filip Hanzely, Peter Richt'arik2019 · 4 citationsarXiv
Distributed Learning with Compressed Gradient Differences1901.09269v3 · Konstantin Mishchenko, Eduard Gorbunov, Martin Tak'ac et al.2019 · 115 citationsarXiv
A Stochastic Penalty Model for Convex and Nonconvex Optimization with Big Constraints1810.13387v1 · Konstantin Mishchenko, Peter Richt'arik2018 · 3 citationsarXiv
SEGA: Variance Reduction via Gradient Sketching1809.03054v2 · Filip Hanzely, Konstantin Mishchenko, Peter Richtarik2018 · 37 citationsarXiv
A Distributed Flexible Delay-tolerant Proximal Gradient Algorithm1806.09429v3 · Konstantin Mishchenko, Franck Iutzeler, J'erome Malick2018 · 21 citationsarXiv
Career total: 51 works. 38 are in this corpus.
Profile built from the corpus for this byline.
Author records are still filling in while the Hub is in alpha. If this is your page, you'll be able to claim it soon. Spot a mistake? Tell us.