{"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/efficient-decentralized-deep-learning-by","title":"Efficient Decentralized Deep Learning by Dynamic Model Averaging","arxiv_id":"1807.03210","date":"2018-07-09","proceeding":null,"authors":["Michael Kamp","Linara Adilova","Joachim Sicking","Fabian Hüger","Peter Schlicht","Tim Wirtz","Stefan Wrobel"],"abstract":"We propose an efficient protocol for decentralized training of deep neural\nnetworks from distributed data sources. The proposed protocol allows to handle\ndifferent phases of model training equally well and to quickly adapt to concept\ndrifts. This leads to a reduction of communication by an order of magnitude\ncompared to periodically communicating state-of-the-art approaches. Moreover,\nwe derive a communication bound that scales well with the hardness of the\nserialized learning problem. The reduction in communication comes at almost no\ncost, as the predictive performance remains virtually unchanged. Indeed, the\nproposed protocol retains loss bounds of periodically averaging schemes. An\nextensive empirical evaluation validates major improvement of the trade-off\nbetween model performance and communication which could be beneficial for\nnumerous decentralized learning applications, such as autonomous driving, or\nvoice recognition and image classification on mobile phones.","url_abs":"http://arxiv.org/abs/1807.03210v2","url_pdf":"http://arxiv.org/pdf/1807.03210v2.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":"efficient-decentralized-deep-learning-by","repo_url":"https://github.com/fraunhofer-iais/dlplatform/tree/master/DLplatform","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"autonomous-driving","task_name":"Autonomous Driving"},{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"image-classification","task_name":"image-classification"},{"task_slug":"model","task_name":"model"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1807.03210","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}