The algorithm converges exactly to a stationary point with diminishing steps and reaches a neighborhood at sublinear rate with fixed steps under stochastic zeroth-order information and compression.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.OC 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Compressed Momentum-based Single-Point Zeroth-Order Algorithm for Stochastic Distributed Nonconvex Optimization
The algorithm converges exactly to a stationary point with diminishing steps and reaches a neighborhood at sublinear rate with fixed steps under stochastic zeroth-order information and compression.