arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2202.01268cs.LG
收藏

DASHA:具有通信压缩、最佳Oracle复杂性和无客户端同步的分布式非凸优化

DASHA: Distributed Nonconvex Optimization with Communication Compression, Optimal Oracle Complexity, and No Client Synchronization

KAUST(阿卜杜拉国王科技大学)

Alexander Tyurin, Peter Richtárik

摘要:

我们开发和分析DASHA:非凸分布优化问题的一个新的家庭的方法。当节点处的局部函数具有有限和或期望形式时,我们的新方法DASHA-PAGE和DASHA-SYNC-MVR改进了Gorbunov等人(2020)先前最先进方法MARINA的理论预言和通信复杂性。特别是,为了实现一个ε-平稳点,并考虑随机稀疏RandK作为一个例子,我们的方法计算的最佳数量的梯度$\mathcal{O}\left(\frac{\sqrt{m}}{\varepsilon\sqrt{n}}\right)$和$\mathcal{O}\left(\fracσ{\varepsilon^{3/2}n}\right)$在有限和和期望形式的情况下,分别,同时保持SOTA通信复杂性$\mathcal{O}\left(\frac{d}{\varepsilon \sqrt{n}}\right)$。此外,与MARINA不同,新方法DASHA、DASHA-PAGE和DASHA-MVR只发送压缩向量,从不同步节点,这使得它们对于联邦学习更实用。我们将我们的结果推广到函数满足Polyak-Lojasiewicz条件的情况。最后,我们的理论在实践中得到了证实:我们在非凸分类和深度学习模型训练的实验中看到了显着的改进。

英文摘要:

We develop and analyze DASHA: a new family of methods for nonconvex distributed optimization problems. When the local functions at the nodes have a finite-sum or an expectation form, our new methods, DASHA-PAGE and DASHA-SYNC-MVR, improve the theoretical oracle and communication complexity of the previous state-of-the-art method MARINA by Gorbunov et al. (2020). In particular, to achieve an epsilon-stationary point, and considering the random sparsifier RandK as an example, our methods compute the optimal number of gradients $\mathcal{O}\left(\frac{\sqrt{m}}{\varepsilon\sqrt{n}}\right)$ and $\mathcal{O}\left(\fracσ{\varepsilon^{3/2}n}\right)$ in finite-sum and expectation form cases, respectively, while maintaining the SOTA communication complexity $\mathcal{O}\left(\frac{d}{\varepsilon \sqrt{n}}\right)$. Furthermore, unlike MARINA, the new methods DASHA, DASHA-PAGE and DASHA-MVR send compressed vectors only and never synchronize the nodes, which makes them more practical for federated learning. We extend our results to the case when the functions satisfy the Polyak-Lojasiewicz condition. Finally, our theory is corroborated in practice: we see a significant improvement in experiments with nonconvex classification and training of deep learning models.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑