{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2024,8,19]],"date-time":"2024-08-19T19:51:44Z","timestamp":1724097104717},"reference-count":0,"publisher":"Association for the Advancement of Artificial Intelligence (AAAI)","issue":"11","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["AAAI"],"abstract":"The stability and generalization of stochastic gradient-based\nmethods provide valuable insights into understanding the algorithmic performance of machine learning models. As the\nmain workhorse for deep learning, the stochastic gradient descent has received a considerable amount of studies. Nevertheless, the community paid little attention to its decentralized variants. In this paper, we provide a novel formulation\nof the decentralized stochastic gradient descent. Leveraging\nthis formulation together with (non)convex optimization theory, we establish the first stability and generalization guarantees for the decentralized stochastic gradient descent. Our\ntheoretical results are built on top of a few common and mild\nassumptions and reveal that the decentralization deteriorates\nthe stability of SGD for the first time. We verify our theoretical findings by using a variety of decentralized settings and\nbenchmark machine learning models.<\/jats:p>","DOI":"10.1609\/aaai.v35i11.17173","type":"journal-article","created":{"date-parts":[[2022,9,8]],"date-time":"2022-09-08T19:26:39Z","timestamp":1662665199000},"page":"9756-9764","source":"Crossref","is-referenced-by-count":7,"title":["Stability and Generalization of Decentralized Stochastic Gradient Descent"],"prefix":"10.1609","volume":"35","author":[{"given":"Tao","family":"Sun","sequence":"first","affiliation":[]},{"given":"Dongsheng","family":"Li","sequence":"additional","affiliation":[]},{"given":"Bao","family":"Wang","sequence":"additional","affiliation":[]}],"member":"9382","published-online":{"date-parts":[[2021,5,18]]},"container-title":["Proceedings of the AAAI Conference on Artificial Intelligence"],"original-title":[],"link":[{"URL":"https:\/\/ojs.aaai.org\/index.php\/AAAI\/article\/download\/17173\/16980","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/ojs.aaai.org\/index.php\/AAAI\/article\/download\/17173\/16980","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2022,9,8]],"date-time":"2022-09-08T19:26:40Z","timestamp":1662665200000},"score":1,"resource":{"primary":{"URL":"https:\/\/ojs.aaai.org\/index.php\/AAAI\/article\/view\/17173"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2021,5,18]]},"references-count":0,"journal-issue":{"issue":"11","published-online":{"date-parts":[[2021,5,28]]}},"URL":"https:\/\/doi.org\/10.1609\/aaai.v35i11.17173","relation":{},"ISSN":["2374-3468","2159-5399"],"issn-type":[{"value":"2374-3468","type":"electronic"},{"value":"2159-5399","type":"print"}],"subject":[],"published":{"date-parts":[[2021,5,18]]}}}