{"corpus_id":6628106,"paper_sha":"a6cb366736791bcccc5c8639de5a8f9636bf87e8","doi":null,"arxiv_id":"1412.6980","pmid":null,"pmcid":null,"mag_id":2953006900,"dblp_id":"journals/corr/KingmaB14","acl_id":null,"title":"Adam: A Method for Stochastic Optimization","year":2014,"publication_date":"2014-12-22","venue":"International Conference on Learning Representations","journal":{"name":"CoRR","pages":null,"volume":"abs/1412.6980"},"journal_issn":null,"journal_title":null,"publication_types":["JournalArticle"],"pubmed_pub_types":null,"s2_fields_of_study":["Mathematics","Computer Science"],"reference_count":26,"citation_count":163577,"influential_citation_count":26003,"is_open_access":false,"arxiv_categories":["cs.LG"],"arxiv_license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","arxiv_journal_ref":null,"mesh_headings":null,"chemicals":null,"comments_corrections":null,"source_flags":1,"s2_open_access_pdf_url":null,"s2_open_access_landing_url":null,"s2_open_access_license":null,"s2_open_access_status":null,"pmc_open_access_pdf_url":null,"pmc_open_access_landing_url":null,"pmc_open_access_license":null,"pmc_open_access_status":null,"unpaywall_open_access_pdf_url":null,"unpaywall_open_access_landing_url":null,"unpaywall_open_access_license":null,"unpaywall_open_access_status":null,"abstract":"We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments. The method is straightforward to implement, is computationally efficient, has little memory requirements, is invariant to diagonal rescaling of the gradients, and is well suited for problems that are large in terms of data and/or parameters. The method is also appropriate for non-stationary objectives and problems with very noisy and/or sparse gradients. The hyper-parameters have intuitive interpretations and typically require little tuning. Some connections to related algorithms, on which Adam was inspired, are discussed. We also analyze the theoretical convergence properties of the algorithm and provide a regret bound on the convergence rate that is comparable to the best known results under the online convex optimization framework. Empirical results demonstrate that Adam works well in practice and compares favorably to other stochastic optimization methods. Finally, we discuss AdaMax, a variant of Adam based on the infinity norm.","claims":[{"public_id":"cl_ad4c106b21432adf95a469f6cddf5a3e","status":"active","text":"AdaMax is a variant of Adam based on the infinity norm.","confidence":0.99,"contributors":[],"url":"https://sah.borca.ai/claims/cl_ad4c106b21432adf95a469f6cddf5a3e"},{"public_id":"cl_e6f776d953f17a04efec3fc4574486ea","status":"active","text":"Adam compares favorably to other stochastic optimization methods in empirical evaluations.","confidence":0.95,"contributors":[],"url":"https://sah.borca.ai/claims/cl_e6f776d953f17a04efec3fc4574486ea"},{"public_id":"cl_efb42f41ea4201c1c99d1971c76e7cbe","status":"active","text":"Adam is computationally efficient, has little memory requirements, is invariant to diagonal rescaling of gradients, and is well suited for problems that are large in terms of data and/or parameters.","confidence":0.98,"contributors":[],"url":"https://sah.borca.ai/claims/cl_efb42f41ea4201c1c99d1971c76e7cbe"},{"public_id":"cl_c70a71781ed60184ba5673ee202ecb4e","status":"active","text":"Adam provides a regret bound on the convergence rate that is comparable to the best known results under the online convex optimization framework.","confidence":0.95,"contributors":[],"url":"https://sah.borca.ai/claims/cl_c70a71781ed60184ba5673ee202ecb4e"}],"concepts":[{"public_id":"co_37f289ef2113335dc5880e2fd92fd8d4","status":"active","name":"adamax","description":"","types":[],"aliases":[],"contributors":[],"url":"https://sah.borca.ai/concepts/co_37f289ef2113335dc5880e2fd92fd8d4"},{"public_id":"co_4a6d5898ae21494f5025375f2f13edc2","status":"active","name":"stochastic optimization","description":"Optimization methods used for objective functions that involve random or stochastic elements.","types":[],"aliases":[],"contributors":[],"url":"https://sah.borca.ai/concepts/co_4a6d5898ae21494f5025375f2f13edc2"},{"public_id":"co_68bb5540c19553a90865d67d83bd07a4","status":"active","name":"regret bound","description":"A theoretical bound on the convergence rate of an algorithm within an online optimization setting.","types":[],"aliases":[],"contributors":[],"url":"https://sah.borca.ai/concepts/co_68bb5540c19553a90865d67d83bd07a4"},{"public_id":"co_a9c9a0223c65faf4c8bd083805eb0f31","status":"active","name":"adam","description":"An algorithm for first-order gradient-based optimization of stochastic objective functions based on adaptive estimates of lower-order moments.","types":[],"aliases":[],"contributors":[],"url":"https://sah.borca.ai/concepts/co_a9c9a0223c65faf4c8bd083805eb0f31"},{"public_id":"co_ac9b1e6877500ad13c74a0279abd7815","status":"active","name":"online convex optimization","description":"A sequential decision-making setting for online learning in which an algorithm repeatedly chooses convex actions or predictions and suffers convex losses, often analyzed via regret.","types":[],"aliases":[],"contributors":[],"url":"https://sah.borca.ai/concepts/co_ac9b1e6877500ad13c74a0279abd7815"}],"external_ids":{"DOI":null,"ArXiv":"1412.6980","PubMed":null,"PubMedCentral":null,"MAG":2953006900,"DBLP":"journals/corr/KingmaB14","ACL":null},"open_access":{"is_open_access":true,"pdf_url":"https://arxiv.org/pdf/1412.6980","landing_url":"https://arxiv.org/abs/1412.6980","source":"arxiv","pdf_url_source":"derived_arxiv","license":"http://arxiv.org/licenses/nonexclusive-distrib/1.0/","reason":null},"reference_availability":{"status":"available","references_indexed":true,"full_text_available":true,"full_text_source":"arxiv","count_basis":"semantic_scholar_metadata","extraction_status":"not_applicable","reason":null},"source":{"provider":"episteme2","base_corpus":"semantic_scholar_dump","freshness_mode":"unknown","basis":["semantic_scholar_metadata","postgres_metadata"],"limits":["paper metadata is based on indexed upstream scholarly datasets","claims and concepts are available only for extracted papers","absence of claims or concepts means no extracted graph data is available in this response"],"status":"available","degraded":false,"degraded_reasons":[],"diagnostics":{"status":"available","degraded":false,"degraded_reasons":[],"metadata_status":"available","graph_status":"available","abstract_status":"available"},"source_flags":1},"paper_id":23795,"paper_uid":"947f181c-2514-45ff-a030-063e5c48afe0","canonical_identity":{"paper_id":23795,"paper_uid":"947f181c-2514-45ff-a030-063e5c48afe0","identity_status":"available","lookup_basis":"semantic_scholar_external_id","compatibility_path":"corpus_id"},"url":"https://sah.borca.ai/papers/6628106"}