
The authors propose two communication-efficient decentralized optimization algorithms, Compressed Push-Pull (CPP) and Broadcast-like CPP (B-CPP), that achieve linear convergence for minimizing strongly convex and smooth objective functions over general directed networks.


coremsg

compressed-gradient-tracking-for-decentralized-optimization-over-general-directed-networks


Compressed Gradient Tracking for Decentralized Optimization Over General Directed Networks


title_rewrite


The authors propose a new projection-free decentralized optimization method, the Inexact Primal-Dual Sliding (I-PDS) algorithm, that achieves communication efficiency and reduces the number of data oracle calls compared to prior work.


efficient-decentralized-optimization-for-machine-learning-with-reduced-data-access


Efficient Decentralized Optimization for Machine Learning with Reduced Data Access
