{"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/bridging-the-gap-between-stochastic-gradient","title":"Bridging the Gap between Stochastic Gradient MCMC and Stochastic Optimization","arxiv_id":"1512.07962","date":"2015-12-25","proceeding":null,"authors":["Changyou Chen","David Carlson","Zhe Gan","Chunyuan Li","Lawrence Carin"],"abstract":"Stochastic gradient Markov chain Monte Carlo (SG-MCMC) methods are Bayesian\nanalogs to popular stochastic optimization methods; however, this connection is\nnot well studied. We explore this relationship by applying simulated annealing\nto an SGMCMC algorithm. Furthermore, we extend recent SG-MCMC methods with two\nkey components: i) adaptive preconditioners (as in ADAgrad or RMSprop), and ii)\nadaptive element-wise momentum weights. The zero-temperature limit gives a\nnovel stochastic optimization method with adaptive element-wise momentum\nweights, while conventional optimization methods only have a shared, static\nmomentum weight. Under certain assumptions, our theoretical analysis suggests\nthe proposed simulated annealing approach converges close to the global optima.\nExperiments on several deep neural network models show state-of-the-art results\ncompared to related stochastic optimization algorithms.","url_abs":"http://arxiv.org/abs/1512.07962v3","url_pdf":"http://arxiv.org/pdf/1512.07962v3.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":"bridging-the-gap-between-stochastic-gradient","repo_url":"https://github.com/cchangyou/Santa","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"stochastic-optimization","task_name":"Stochastic Optimization"}],"methods":[{"method_slug":"adagrad","method_name":"AdaGrad"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1512.07962","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}