Bilevel optimization has gained significant attention in recent years because of its broad applications in machine learning. This paper focuses on bilevel optimization in decentralized networks and proposes a novel single-loop algorithm for solving decentralized bilevel optimization with a strongly convex lower-level problem. Our approach is built on the basis of the SOBA framework, and it is a fully single-loop method that approximates the hypergradient by using merely two matrix-vector multiplications per iteration. Importantly, by incorporating the gradient tracking and projection techniques, our algorithm does not require any gradient heterogeneity assumption, which distinguishes it from existing methods for decentralized bilevel optimization and federated bilevel optimization. We establish the convergence rate of the proposed algorithm. Moreover, we present experimental results on hyperparameter optimization and data hyper-cleaning problems, which demonstrate the efficiency of our proposed algorithm. Funding: This work was supported by the National Science Foundation Division of Mathematical Sciences [Grant 2243650]; the Division of Computing and Communication Foundations [Grants 2308597 and 2311275]; the Division of Electrical, Communications and Cyber Systems [Grant 2326591]; the National Natural Science Foundation of China [Grants 12431011, 12371301 and 12501425]; the Key Laboratory of NSLSCS, Ministry of Education of China; and the Natural Science Foundation of Jiangsu Province [Grant SBK20250404650].
more »
« less
This content will become publicly available on May 1, 2027
On the Communication Complexity of Decentralized Stochastic Bilevel Optimization
Abstract Stochastic bilevel optimization finds widespread applications in machine learning, including meta-learning, hyperparameter optimization, and neural architecture search. To extend stochastic bilevel optimization to distributed data, several decentralized stochastic bilevel optimization algorithms have been developed. However, existing methods often suffer from slow convergence rates and high communication costs in heterogeneous settings, limiting their applicability to real-world tasks. To address these issues, we propose two novel decentralized stochastic bilevel gradient descent algorithms based on simultaneous and alternating update strategies. Our algorithms can achieve faster convergence rates and lower communication costs than existing methods. Importantly, our convergence analyses do not rely on strong assumptions regarding heterogeneity. More importantly, our theoretical analyses clearly disclose how the computation and communication regarding the Hessian-inverse-vector product under the heterogeneous setting affects the convergence rate. To the best of our knowledge, this is the first time such favorable theoretical results have been achieved with mild assumptions in the heterogeneous setting. Furthermore, we demonstrate how to establish the convergence rate for the alternating update strategy when combined with the variance-reduced gradient. Finally, experimental results confirm the efficacy of our algorithms.
more »
« less
- PAR ID:
- 10704428
- Publisher / Repository:
- Springer
- Date Published:
- Journal Name:
- Machine Learning
- Volume:
- 115
- Issue:
- 5
- ISSN:
- 0885-6125
- Subject(s) / Keyword(s):
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Standard federated optimization methods successfully apply to stochastic problems with singlelevel structure. However, many contemporary ML problems – including adversarial robustness, hyperparameter tuning, actor-critic – fall under nested bilevel programming that subsumes minimax and compositional optimization. In this work, we propose FEDNEST: A federated alternating stochastic gradient method to address general nested problems. We establish provable convergence rates for FEDNEST in the presence of heterogeneous data and introduce variations for bilevel, minimax, and compositional optimization. FEDNEST introduces multiple innovations including federated hypergradient computation and variance reduction to address inner-level heterogeneity. We complement our theory with experiments on hyperparameter & hyper-representation learning and minimax optimization that demonstrate the benefits of our method in practice.more » « less
-
Federated stochastic bilevel optimization has been actively studied in recent years due to its widespread applications in machine learning. However, most existing federated stochastic bilevel optimization algorithms require the computation of second-order Hessian and Jacobian matrices, which leads to longer running times in practice. To address these challenges, we propose a novel federated stochastic variance-reduced bilevel gradient descent algorithm that relies solely on first-order oracles. Specifically, our approach does not require the computation of second-order Hessian and Jacobian matrices, significantly reducing running time. Furthermore, we introduce a novel learning rate mechanism, i.e., a constant single-time-scale learning rate, to coordinate the update of different variables. We also present a new strategy to establish the convergence rate of our algorithm. Finally, the extensive experimental results confirm the efficacy of our proposed algorithm.more » « less
-
Stochastic gradient descent (SGD) is one of the most widely used optimization methods for parallel and distributed processing of large datasets. One of the key limitations of distributed SGD is the need to regularly communicate the gradients between different computation nodes. To reduce this communication bottleneck, recent work has considered a one-bit variant of SGD, where only the sign of each gradient element is used in optimization. In this paper, we extend this idea by proposing a stochastic variant of the proximal-gradient method that also uses one-bit per update element. We prove the theoretical convergence of the method for non-convex optimization under a set of explicit assumptions. Our results indicate that the compressed method can match the convergence rate of the uncompressed one, making the proposed method potentially appealing for distributed processing of large datasets.more » « less
-
Communication compression has become a key strategy to speed up distributed optimization. However, existing decentralized algorithms with compression mainly focus on compressing DGD-type algorithms. They are unsatisfactory in terms of convergence rate, stability, and the capability to handle heterogeneous data. Motivated by primal-dual algorithms, this paper proposes the first \underline{L}in\underline{EA}r convergent \underline{D}ecentralized algorithm with compression, LEAD. Our theory describes the coupled dynamics of the inexact primal and dual update as well as compression error, and we provide the first consensus error bound in such settings without assuming bounded gradients. Experiments on convex problems validate our theoretical analysis, and empirical study on deep neural nets shows that LEAD is applicable to non-convex problems.more » « less
An official website of the United States government
