LGMLJun 25, 2022

Topology-aware Generalization of Decentralized SGD

arXiv:2206.12680v559 citationsh-index: 50Has Code
Originality Incremental advance
AI Analysis

It addresses the generalization performance of decentralized learning algorithms, which is crucial for distributed machine learning systems, but is incremental as it builds on existing D-SGD analysis by focusing on topology-aware aspects.

This paper tackles the problem of understanding the generalization ability of decentralized stochastic gradient descent (D-SGD) by proving stability and generalization bounds that depend on the communication topology's spectral gap, showing that better connectivity improves generalization, with experiments on image datasets validating the theory.

This paper studies the algorithmic stability and generalizability of decentralized stochastic gradient descent (D-SGD). We prove that the consensus model learned by D-SGD is $\mathcal{O}{(N^{-1}+m^{-1} +λ^2)}$-stable in expectation in the non-convex non-smooth setting, where $N$ is the total sample size, $m$ is the worker number, and $1+λ$ is the spectral gap that measures the connectivity of the communication topology. These results then deliver an $\mathcal{O}{(N^{-(1+α)/2}+ m^{-(1+α)/2}+λ^{1+α} + φ_{\mathcal{S}})}$ in-average generalization bound, which is non-vacuous even when $λ$ is closed to $1$, in contrast to vacuous as suggested by existing literature on the projected version of D-SGD. Our theory indicates that the generalizability of D-SGD is positively correlated with the spectral gap, and can explain why consensus control in initial training phase can ensure better generalization. Experiments of VGG-11 and ResNet-18 on CIFAR-10, CIFAR-100 and Tiny-ImageNet justify our theory. To our best knowledge, this is the first work on the topology-aware generalization of vanilla D-SGD. Code is available at https://github.com/Raiden-Zhu/Generalization-of-DSGD.

Code Implementations1 repo
Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes