{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/adding-vs-averaging-in-distributed-primal","title":"Adding vs. Averaging in Distributed Primal-Dual Optimization","arxiv_id":"1502.03508","date":"2015-02-12","proceeding":null,"authors":["Chenxin Ma","Virginia Smith","Martin Jaggi","Michael. I. Jordan","Peter Richtárik","Martin Takáč"],"abstract":"Distributed optimization methods for large-scale machine learning suffer from\na communication bottleneck. It is difficult to reduce this bottleneck while\nstill efficiently and accurately aggregating partial work from different\nmachines. In this paper, we present a novel generalization of the recent\ncommunication-efficient primal-dual framework (CoCoA) for distributed\noptimization. Our framework, CoCoA+, allows for additive combination of local\nupdates to the global parameters at each iteration, whereas previous schemes\nwith convergence guarantees only allow conservative averaging. We give stronger\n(primal-dual) convergence rate guarantees for both CoCoA as well as our new\nvariants, and generalize the theory for both methods to cover non-smooth convex\nloss functions. We provide an extensive experimental comparison that shows the\nmarkedly improved performance of CoCoA+ on several real-world distributed\ndatasets, especially when scaling up the number of machines.","url_abs":"http://arxiv.org/abs/1502.03508v2","url_pdf":"http://arxiv.org/pdf/1502.03508v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"adding-vs-averaging-in-distributed-primal","repo_url":"https://github.com/gingsmith/cocoa","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"distributed-optimization","task_name":"Distributed Optimization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1502.03508","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}