Peter Richtarik

Latest

• Hyperparameter Transfer Learning with Adaptive Complexity
• Optimal Client Sampling for Federated Learning
• Lower Bounds and Optimal Algorithms for Personalized Federated Learning
• A Better Alternative to Error Feedback for Communication-Efficient Distributed Learning
• On Biased Compression for Distributed Learning
• Adaptivity of Stochastic Gradient Methods for Nonconvex Optimization
• Natural Compression for Distributed Deep Learning
• Nonconvex Variance Reduced Optimization with Arbitrary Sampling
• Stochastic Distributed Learning with Gradient Quantization and Variance Reduction
• Don’t Jump Through Hoops and Remove Those Loops: SVRG and Katyusha are Better Without the Outer Loop