Abstract
Classically, the design of multi-agent systems is approached using techniques from distributed optimization such as dual descent and consensus algorithms. Such algorithms depend on convergence to global consensus before any individual agent can determine its local action. This leads to challenges with respect to communication overhead and robustness, and improving algorithms with respect to these measures has been a focus of the community for decades.
This paper presents a new approach for multi-agent system design based on ideas from the emerging field of local computation algorithms. The framework we develop, LOcal Convex Optimization (LOCO), is the first local computation algorithm for convex optimization problems and can be applied in a wide-variety of settings. We demonstrate the generality of the framework via applications to Network Utility Maximization (NUM) and the distributed training of Support Vector Machines (SVMs), providing numerical results illustrating the improvement compared to classical distributed optimization approaches in each case.
- Dimitris Achlioptas, Themis Gouleakis, and Fotis Iliopoulos. 2018. Local Computation Algorithms for the Lová sz Local Lemma. CoRR, Vol. abs/1809.07910 (2018). http://arxiv.org/abs/1809.07910Google Scholar
- Amr Ahmed, Mohamed Aly, Joseph Gonzalez, Shravan Narayanamurthy, and Alexander Smola. 2012. Scalable inference in latent variable models. In Proc. of the 5th ACM WSDM. 123--132.Google Scholar
Digital Library
- Noga Alon, Ronitt Rubinfeld, Shai Vardi, and Ning Xie. 2012. Space-Efficient Local Computation Algorithms. In Proc. 22ndACM-SIAM Symposium on Discrete Algorithms (SODA). 1132--1139.Google Scholar
Cross Ref
- Ganesh Ananthanarayanan, Michael Chien-Chun Hung, Xiaoqi Ren, Ion Stoica, Adam Wierman, and Minlan Yu. 2014. GRASS: Trimming Stragglers in Approximation Analytics. In 11th USENIX Symposium on Networked Systems Design and Implementation (NSDI 14).Google Scholar
Digital Library
- Reid Andersen, ChristianBorgs, Jennifer Chayes, John Hopcraft, Vahab S. Mirrokni, and Shang-Hua Teng. 2008. Local Computation of PageRank Contributions. Internet Mathematics, Vol. 5(1--2) (2008), 23--45.Google Scholar
Cross Ref
- Yossi Azar, Niv Buchbinder, T-H. Hubert Chan, Shahar Chen, Ilan Reuven Cohen, Anupam Gupta, Zhiyi Huang, Ning Kang, Viswanath Nagarajan, Joseph (Seffi) Naor, and Debmalya Panigrahi. 2016. Online algorithms for covering and packing problems with convex objectives. In Proc. of the IEEE 57th Annual Symposium on Foundations of Computer Science (FOCS). 148--157.Google Scholar
Cross Ref
- Amrit Singh Bedi and Ketan Rajawat. 2017. Asynchronous Incremental Stochastic Dual Descent Algorithm for Network Resource Allocation. IEEE Transactions on Signal Processing, Vol. 66 (2017), 2229--2244.Google Scholar
Cross Ref
- Jacobus F. Benders. 1962. Partitioning procedures for solving mixed-variables programming problems. Numer. Math., Vol. 4, 1 (1962), 238--252.Google Scholar
Digital Library
- Dimitri P. Bertsekas. 1999. Nonlinear programming .Athena Scientific.Google Scholar
- Dimitri P. Bertsekas and John N. Tsitsiklis. 1989. Parallel and Distributed Computation: Numerical Methods .Prentice Hall.Google Scholar
- Vincent D. Blondel, Julien M. Hendrickx, Alex Olshevsky, and John N. Tsitsiklis. 2005. Convergence in multiagent coordination, consensus, and flocking. In Proc. of IEEE Conference on Decision and Control. 2996--3000.Google Scholar
- Sem Borst, Varun Gupta, and Anwar Walid. 2010. Distributed caching algorithms for content distribution networks. In Proceedings of IEEE INFOCOM. 1--9.Google Scholar
Cross Ref
- Stephen Boyd, Neal Parikh, Eric Chu, Borja Peleato, and Jonathan Eckstein. 2011. Distributed optimization and statistical learning via the alternating direction method of multipliers. Found. and Trends in Machine Learning, Vol. 3, 1 (2011), 1--122.Google Scholar
Digital Library
- Niv Buchbinder and Joseph (Seffi) Naor. 2009. The Design of Competitive Online Algorithms via a Primal-Dual Approach. Foundations and Trends in Theoretical Computer Science, Vol. 3, 2--3 (2009), 93--263.Google Scholar
Digital Library
- CAIDA. 2007., Vol. The CAIDA UCSD AS Relationship Dataset http://www.caida.org/data/as-relationships/ (2007).Google Scholar
- Tsung-Hui Chang, Mingyi Hong, Wei-Cheng Liao, and Xiangfeng Wang. 2016. Asynchronous Distributed ADMM for Large-Scale Optimization-Part I: Algorithm and Convergence Analysis. IEEE Transactions on Signal Processing, Vol. 64, 12 (2016), 3118--3130.Google Scholar
Digital Library
- Nikolaos Chatzipanagiotis, Darinka Dentcheva, and Michael M. Zavlanos. 2015. An Augmented Lagrangian Method for Distributed Optimization. Math. Program., Vol. 152, 1--2 (2015), 405--434.Google Scholar
Digital Library
- Yudong Chen, Lili Su, and Jiaming Xu. 2017. Distributed Statistical Machine Learning in Adversarial Settings: Byzantine Gradient Descent. Proc. ACM Meas. Anal. Comput. Syst. SIGMETRICS, Vol. 1, 2 (2017).Google Scholar
Digital Library
- Mung Chiang, Steven H Low, A Robert Calderbank, and John C Doyle. 2007. Layering as optimization decomposition: A mathematical theory of network architectures. Proc. IEEE, Vol. 95, 1 (2007), 255--312.Google Scholar
Cross Ref
- Patrick L. Combettes and Valerie R. Wajs. 2005. Signal recovery by proximal forward-backward splitting. Multiscale Modeling & Simulation, Vol. 4, 4 (2005), 1168--1200.Google Scholar
Cross Ref
- Corinna Cortes and Vladimir Vapnik. 1995. Support-vector networks. Machine Learning, Vol. 20, 3 (1995), 273--297.Google Scholar
Cross Ref
- George Dantzig. 2016. Linear programming and extensions .Princeton university press.Google Scholar
- George B. Dantzig and Philip Wolfe. 1960. Decomposition principle for linear programs. Oper. Res., Vol. 8, 1 (1960), 101--111.Google Scholar
Digital Library
- Reza Eghbali and Maryam Fazel. 2016. Designing smoothing functions for improved worst-case competitive ratio in online optimization. In Neural Information Processing Systems. 3287--3295.Google Scholar
- Tomaso Erseghe. 2014. Distributed optimal power flow using ADMM. IEEE Trans. on Power Sys., Vol. 29, 5 (2014), 2370--2380.Google Scholar
Cross Ref
- Tomaso Erseghe, Davide Zennaro, Emiliano Dall'Anese, and Lorenzo Vangelista. 2011. Fast Consensus by the Alternating Direction Multipliers Method. IEEE Trans. on Sig. Proces., Vol. 59 (2011), 5523--5537.Google Scholar
Digital Library
- Uriel Feige, Boaz Patt-Shamir, and Shai Vardi. 2018. On the Probe Complexity of Local Computation Algorithms. In 45th International Colloquium on Automata, Languages, and Programming, (ICALP). 50:1--50:14.Google Scholar
- Pedro A Forero, Alfonso Cano, and Georgios B. Giannakis. 2010. Consensus-Based Distributed Support Vector Machines. J. Mach. Learn. Res., Vol. 11 (2010), 1663--1707.Google Scholar
Digital Library
- Daniel Gabay and Bertrand Mercier. 1976. A dual algorithm for the solution of nonlinear variational problems via finite element approximation. Computers & Mathematics with Applications, Vol. 2, 1 (1976), 17--40.Google Scholar
Cross Ref
- Lingwen Gan, Ufuk Topcu, and Steven H. Low. 2013. Optimal decentralized protocol for electric vehicle charging. IEEE Transactions on Power Systems, Vol. 28, 2 (2013), 940--951.Google Scholar
Cross Ref
- Tom Goldstein, Gavin Taylor, Kawika Barabin, and Kent Sayre. 2016. Unwrapping ADMM: Efficient Distributed Computing via Transpose Reduction. In AISTATS. 1151--1158.Google Scholar
- Joseph E. Gonzalez, Yucheng Low, Carlos E. Guestrin, and David O'Hallaron. 2009. Distributed parallel inference on large factor graphs. In Proc. of the 25th Conf. on UIAI. 203--212.Google Scholar
- Carlos Guestrin, Peter Bodik, Romain Thibaux, Mark Paskin, and Samuel Madden. 2004. Distributed regression: an efficient framework for modeling sensor network data. In Proceedings of the 3rd ACM IPSN. 1--10.Google Scholar
Digital Library
- Yi Guo and Lynne E Parker. 2002. A distributed and optimal motion planning approach for multiple mobile robots. In Robotics and Automation, 2002. Proceedings. ICRA'02. IEEE International Conference on, Vol. 3. 2612--2619.Google Scholar
- Tamir Hazan, Amit Man, and Amnon Shashua. 2008. A parallel decomposition solver for SVM: Distributed dual ascend using fenchel duality. In Proc. of CVPR. 1--8.Google Scholar
Cross Ref
- Jianghai Hu, Yingying Xiao, and Ji Liu. 2018. Distributed Algorithms for Solving Locally Coupled Optimization Problems on Agent Networks. In Decision and Control (CDC), 2007 IEEE Annual Conference on. IEEE, 2420--2425.Google Scholar
Cross Ref
- Hugh Everett III. 1963. Generalized Lagrange multiplier method for solving problems of optimum allocation of resources. Operations research, Vol. 11, 3 (1963), 399--417.Google Scholar
- Thorsten Joachims. 2006. Training Linear SVMs in Linear Time. In Proceedings of the 12th ACM SIGKDD. 217--226.Google Scholar
Digital Library
- Jonathan Katz and Luca Trevisan. 2000. On the efficiency of local decoding procedures for error-correcting codes. In Proc. 32nd Annual ACM Symposium on the Theory of Computing (STOC). 80--86.Google Scholar
Digital Library
- Frank P. Kelly, A. K. Maulloo, and David K. H. Tan. 1998. Rate control for communication networks: shadow prices, proportional fairness and stability. J. of the Operational Research Society, Vol. 49, 3 (1998), 237--252.Google Scholar
Cross Ref
- Sarit Khirirat, Mikael Johansson, and Dan Alistarh. 2018. Gradient compression for communication-limited convex optimization. In 2018 IEEE Conference on Decision and Control (CDC). 166--171.Google Scholar
Cross Ref
- Jakub Konecný, H. Brendan McMahan, and Daniel Ramage. 2015. Federated Optimization: Distributed Optimization Beyond the Datacenter. ArXiv, Vol. abs/1511.03575 (2015).Google Scholar
- Christos Koufogiannakis and Neal E. Young. 2011. Distributed algorithms for covering, packing and maximum weighted matching. Distributed Computing, Vol. 24, 1 (2011), 45--63.Google Scholar
Digital Library
- Yoshiaki Kuwata and Jonathan P How. 2011. Cooperative distributed robust trajectory optimization using receding horizon MILP. IEEE Transactions on Control Systems Technology, Vol. 19, 2 (2011), 423--431.Google Scholar
Cross Ref
- Leon S Lasdon. 1970. Optimization theory for large systems .Courier Corporation.Google Scholar
- Reut Levi and Ronitt Rubinfeld andAnak Yodpinyanee. 2015. Brief Announcement: Local Computation Algorithms for Graphs of Non-Constant Degrees. In Proc. of the 27th ACM on Symposium on Parallelism in Algorithms and Architectures, (SPAA). 59--61.Google Scholar
Digital Library
- David D. Lewis, Yiming Yang, Tony G. Rose, and Fan Li. 2004. RCV1: A New Benchmark Collection for Text Categorization Research. Journal of Machine Learning Research., Vol. 5 (2004), 361--397.Google Scholar
Digital Library
- Na Li, Lijun Chen, and Steven H Low. 2011. Optimal demand response based on utility maximization in power networks. In IEEE Power and Energy Society General Meeting. 1--8.Google Scholar
Cross Ref
- Ying Liao, Huan Qi, and Weiqun Li. 2013. Load-balanced clustering algorithm with distributed self-organization for wireless sensor networks. IEEE Sensors Journal, Vol. 13, 5 (2013), 1498--1506.Google Scholar
Cross Ref
- Steven H. Low and David E. Lapsley. 1999. Optimization flow control. I. Basic algorithm and convergence. IEEE/ACM Transactions on Networking, Vol. 7, 6 (1999), 861--874.Google Scholar
Digital Library
- Steven H Low, Fernando Paganini, and John C Doyle. 2002. Internet congestion control. IEEE Control Systems, Vol. 22, 1 (2002), 28--43.Google Scholar
Cross Ref
- Sindri Magnússon, Chinwendu Enyioha, Na Li, Carlo Fischione, and Vahid Tarokh. 2018a. Communication Complexity of Dual Decomposition Methods for Distributed Resource Allocation Optimization. IEEE Journal of Selected Topics in Signal Processing, Vol. 12, 4 (2018), 717--732.Google Scholar
Cross Ref
- Sindri Magnússon, Chinwendu Enyioha, Na Li, Carlo Fischione, and Vahid Tarokh. 2018b. Convergence of limited communication gradient methods. IEEE Journal of Selected Topics in Signal Processing, Vol. 63, 5 (2018), 1356--1371.Google Scholar
- Yishay Mansour, Aviad Rubinstein, Shai Vardi, and Ning Xie. 2012. Converting Online Algorithms to Local Computation Algorithms. In Proc. of 39th Intl. Colloq. on Automata, Lang. and Prog. (ICALP). 653--664.Google Scholar
Digital Library
- Laurent Massoulié and James Roberts. 1999. Bandwidth sharing: objectives and algorithms. In IEEE INFOCOM, Vol. 3. 1395--1403.Google Scholar
- Brendan McMahan and Daniel Ramage. Accessed: 2017-04--10. Federated learning: Collaborative machine learning without centralized training data. https:// research.googleblog.com/ 2017/ 04/ federated-l earning-collaborative.html.Google Scholar
- Damon Mosk-Aoyama, Tim Roughgarden, and Devavrat Shah. 2010. Fully distributed algorithms for convex optimization problems. SIAM J. on Opt., Vol. 20, 6 (2010), 3260--3279.Google Scholar
Digital Library
- Jo ao F. C. Mota, Jo ao M. F. Xavier, Pedro M. Q. Aguiar, and Markus Püschel. 2013. D-ADMM: A communication-efficient distributed algorithm for separable optimization. IEEE Trans. on Sig. Proces., Vol. 61, 10 (2013), 2718--2723.Google Scholar
Digital Library
- Angelia Nedic and Asuman Ozdaglar. 2007. On the Rate of Convergence of Distributed Subgradient Methods for Multi-agent Optimization. In Decision and Control (CDC), 2007 IEEE 46th Annual Conference on. IEEE, 4711--4716.Google Scholar
Cross Ref
- Angelia Nedic and Asuman Ozdaglar. 2009. Distributed subgradient methods for multi-agent optimization. IEEE Trans. on Autom. Control, Vol. 54, 1 (2009), 48--61.Google Scholar
Cross Ref
- Angelia Nedić and Asuman Ozdaglar. 2010. Convergence rate for consensus with delays. Journal of Global Optimization, Vol. 47, 3 (2010), 437--456.Google Scholar
Digital Library
- Angelia Nedic, Asuman Ozdaglar, and Pablo A. Parrilo. 2010. Constrained Consensus and Optimization in Multi-Agent Networks. IEEE Trans. Automat. Control, Vol. 55, 4 (2010).Google Scholar
Cross Ref
- Feng Niu, Benjamin Recht, Christopher Re, and Stephen J. Wright. 2011. HOGWILD!: A Lock-free Approach to Parallelizing Stochastic Gradient Descent. In Proceedings of the 24th International Conference on Neural Information Processing Systems (NIPS'11).Google Scholar
Digital Library
- Reza Olfati-Saber. 2007. Distributed Kalman filtering for sensor networks. In Proc. of IEEE CDC. 5492--5498.Google Scholar
Cross Ref
- Venkata N. Padmanabhan, Helen J. Wang, Philip A. Chou, and Kunwadee Sripanidkulchai. 2002. Distributing streaming media content using cooperative networking. In Proceedings of workshop on Network and operating systems support for digital audio and video. ACM, 177--186.Google Scholar
Digital Library
- Daniel P. Palomar and Mung Chiang. 2007. Alternative distributed algorithms for network utility maximization: Framework and applications. IEEE Trans. on Autom. Control, Vol. 52, 12 (2007), 2254--2269.Google Scholar
Cross Ref
- Xinghao Pan, Maximilian Lam, Stephen Tu, Dimitris Papailiopoulos, Ce Zhang, Michael I. Jordan, Kannan Ramchandran, Chris Re, and Benjamin Recht. 2016. CYCLADES: Conflict-free Asynchronous Machine Learning. In Proceedings of the 30th International Conference on Neural Information Processing Systems (NIPS'16).Google Scholar
- Mark A. Paskin, Carlos E. Guestrin, and Jim McFadden. 2005. A robust architecture for distributed inference in sensor networks. In Proceedings of the 4th ACM IPSN.Google Scholar
Cross Ref
- David Peleg. 2000. Distributed Computing: A Locality-Sensitive Approach .SIAM Monographs on Discrete Mathematics and Applications.Google Scholar
Cross Ref
- Qiuyu Peng and Steven H. Low. 2016. Distributed optimal power flow algorithm for radial networks, I: Balanced single phase case. IEEE Transactions on Smart Grid (2016).Google Scholar
- Robin L Raffard, Claire J Tomlin, and Stephen P Boyd. 2004. Distributed optimization for cooperative agents: Application to formation flight. In Proc. of IEEE Conference on Decision and Control, Vol. 3. 2453--2459.Google Scholar
Cross Ref
- Pradeep Ravikumar, Alekh Agarwal, and Martin J. Wainwright. 2010. Message passing for graph-structured linear programs: Proximal methods and rounding schemes. JMLR, Vol. 11 (2010), 1043--1080.Google Scholar
Digital Library
- Omer Reingold and Shai Vardi. 2016. New techniques and tighter bounds for local computation algorithms. Journal of Computer and System Science, Vol. 82, 7 (2016), 1180--1200.Google Scholar
Digital Library
- Ralph Tyrrell Rockafellar. 1984. Network Flows and Monotropic Optimization .John Wiley and Sons, New York.Google Scholar
- Ronitt Rubinfeld, Gil Tamir, Shai Vardi, and Ning Xie. 2011. Fast Local Computation Algorithms. In Proc. 2nd Sym. on Innov. in Computer Science (ICS). 223--238.Google Scholar
- Michael Saks and C. Seshadhri. 2010. Local Monotonicity Reconstruction. SIAM J. on Comp., Vol. 39, 7 (2010), 2897--2926.Google Scholar
Digital Library
- Pedram Samadi, Amir-Hamed Mohsenian-Rad, Robert Schober, Vincent WS Wong, and Juri Jatskevich. 2010. Optimal real-time pricing algorithm based on utility maximization for smart grid. In Proc. of IEEE Smart Grid Communications (SmartGridComm). 415--420.Google Scholar
Cross Ref
- Sujay Sanghavi, Dmitry M. Malioutov, and Alan S. Willsky. 2008. Linear programming analysis of loopy belief propagation for weighted matching. In Proc. of NIPS. 1273--1280.Google Scholar
- Ioannis D. Schizas, Alejandro Ribeiro, and Georgios B. Giannakis. 2008. Consensus in ad hoc WSNs with noisy links-Part I: Distributed estimation of deterministic signals. IEEE Trans. on Signal Processing, Vol. 56, 1 (2008), 350--364.Google Scholar
Digital Library
- Shai Shalev-Shwartz, Yoram Singer, Nathan Srebro, and Andrew Cotter. 2011. Pegasos: primal estimated sub-gradient solver for SVM. Mathematical Programming, Vol. 127, 1 (2011), 3--30.Google Scholar
Digital Library
- Wei Shi, Qing Ling, Gang Wu, and Wotao Yin. 2015. A Proximal Gradient Algorithm for Decentralized Composite Optimization. IEEE Transactions on Signal Processing, Vol. 63, 22 (2015).Google Scholar
Cross Ref
- Naum Z. Shor. 2012. Minimization methods for non-differentiable functions. Vol. 3. Springer Science & Business Media.Google Scholar
- Rayadurgam Srikant. 2012. The mathematics of Internet congestion control .Springer Science & Business Media.Google Scholar
- Gabriele Steidl and Tanja Teuber. 2010. Removing multiplicative noise by Douglas-Rachford splitting methods. Journal of Math. Imaging and Vision, Vol. 36, 2 (2010), 168--184.Google Scholar
Digital Library
- Ichiro Suzuki and Masafumi Yamashita. 1999. Distributed anonymous mobile robots: Formation of geometric patterns. SIAM J. Comput., Vol. 28, 4 (1999), 1347--1363.Google Scholar
Digital Library
- Håkan Terelius, Ufuk Topcu, and Richard M. Murray. 2011. Decentralized multi-agent optimization via dual decomposition. IEEE Trans. Automat. Control, Vol. 44, 1 (2011).Google Scholar
- John N. Tsitsiklis, Dimitri P. Bertsekas, and Michael Athans. 1986. Distributed asynchronous deterministic and stochastic gradient optimization algorithms. IEEE Trans. on Autom. Control, Vol. 31 (1986), 803--812.Google Scholar
Cross Ref
- Ermin Wei, Asuman Ozdaglar, and Ali Jadbabaie. 2015. A distributed Newton method for network utility maximization: Algorithm. IEEE Trans. on Autom. Control, Vol. 58, 9 (2015), 2162--2175.Google Scholar
Cross Ref
- David P. Woodruff. 2014. Sketching as a Tool for Numerical Linear Algebra. Found. and Trends in Theoretical Computer Science, Vol. 10, 1--2 (2014), 1--157.Google Scholar
Digital Library
- Yung Yi and Mung Chiang. 2008. Stochastic network utility maximization -- a tribute to Kelly's paper published in this journal a decade ago. European Transactions on Telecommunications, Vol. 19, 4 (2008), 421--442.Google Scholar
Cross Ref
- Ruiliang Zhang and James T. Kwok. 2014. Asynchronous Distributed ADMM for Consensus Optimization. In Proceedings of the 31st International Conference on International Conference on Machine Learning - Volume 32 (ICML'14). 1701--1709.Google Scholar
Index Terms
Logarithmic Communication for Distributed Optimization in Multi-Agent Systems
Recommendations
Logarithmic Communication for Distributed Optimization in Multi-Agent Systems
SIGMETRICS '20: Abstracts of the 2020 SIGMETRICS/Performance Joint International Conference on Measurement and Modeling of Computer SystemsClassically, the design of multi-agent systems is approached using techniques from distributed optimization such as dual descent and consensus algorithms. Such algorithms depend on convergence to global consensus before any individual agent can ...
Logarithmic Communication for Distributed Optimization in Multi-Agent Systems
Classically, the design of multi-agent systems is approached using techniques from distributed optimization such as dual descent and consensus algorithms. Such algorithms depend on convergence to global consensus before any individual agent can ...
A review of distributed optimization: Problems, models and algorithms
AbstractWith the development of big data and artificial intelligence, distributed optimization has emerged as an indispensable tool for solving large-scale problems. In particular, the multi-agent system based on distributed information ...






Comments