A clipping-based distributed subgradient method is proven to converge for non-smooth weakly convex problems under heavy-tailed noise, with an O(1/log T) stationarity rate.
Machine learning at the wireless edge: Distributed stochastic gradient descent over-the-air
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.OC 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Distributed Stochastic Optimization for Non-Smooth and Weakly Convex Problems under Heavy-Tailed Noise
A clipping-based distributed subgradient method is proven to converge for non-smooth weakly convex problems under heavy-tailed noise, with an O(1/log T) stationarity rate.