We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: DIAMOND: Taming Sample and Communication Complexities in Decentralized Bilevel Optimization

Abstract: Decentralized bilevel optimization has received increasing attention recently due to its foundational role in many emerging multi-agent learning paradigms (e.g., multi-agent meta-learning and multi-agent reinforcement learning) over peer-to-peer edge networks. However, to work with the limited computation and communication capabilities of edge networks, a major challenge in developing decentralized bilevel optimization techniques is to lower sample and communication complexities. This motivates us to develop a new decentralized bilevel optimization called DIAMOND (decentralized single-timescale stochastic approximation with momentum and gradient-tracking). The contributions of this paper are as follows: i) our DIAMOND algorithm adopts a single-loop structure rather than following the natural double-loop structure of bilevel optimization, which offers low computation and implementation complexity; ii) compared to existing approaches, the DIAMOND algorithm does not require any full gradient evaluations, which further reduces both sample and computational complexities; iii) through a careful integration of momentum information and gradient tracking techniques, we show that the DIAMOND algorithm enjoys $\mathcal{O}(\epsilon^{-3/2})$ in sample and communication complexities for achieving an $\epsilon$-stationary solution, both of which are independent of the dataset sizes and significantly outperform existing works. Extensive experiments also verify our theoretical findings.
Subjects: Machine Learning (cs.LG); Optimization and Control (math.OC)
Cite as: arXiv:2212.02376 [cs.LG]
  (or arXiv:2212.02376v4 [cs.LG] for this version)

Submission history

From: Peiwen Qiu [view email]
[v1] Mon, 5 Dec 2022 15:58:00 GMT (4407kb,I)
[v2] Wed, 7 Dec 2022 16:09:28 GMT (4407kb,I)
[v3] Sat, 10 Dec 2022 23:58:44 GMT (4407kb,I)
[v4] Thu, 19 Jan 2023 22:46:09 GMT (4414kb,D)

Link back to: arXiv, form interface, contact.