{"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/hyperadam-a-learnable-task-adaptive-adam-for","title":"HyperAdam: A Learnable Task-Adaptive Adam for Network Training","arxiv_id":"1811.08996","date":"2018-11-22","proceeding":null,"authors":["Shipeng Wang","Jian Sun","Zongben Xu"],"abstract":"Deep neural networks are traditionally trained using human-designed\nstochastic optimization algorithms, such as SGD and Adam. Recently, the\napproach of learning to optimize network parameters has emerged as a promising\nresearch topic. However, these learned black-box optimizers sometimes do not\nfully utilize the experience in human-designed optimizers, therefore have\nlimitation in generalization ability. In this paper, a new optimizer, dubbed as\n\\textit{HyperAdam}, is proposed that combines the idea of \"learning to\noptimize\" and traditional Adam optimizer. Given a network for training, its\nparameter update in each iteration generated by HyperAdam is an adaptive\ncombination of multiple updates generated by Adam with varying decay rates. The\ncombination weights and decay rates in HyperAdam are adaptively learned\ndepending on the task. HyperAdam is modeled as a recurrent neural network with\nAdamCell, WeightCell and StateCell. It is justified to be state-of-the-art for\nvarious network training, such as multilayer perceptron, CNN and LSTM.","url_abs":"http://arxiv.org/abs/1811.08996v1","url_pdf":"http://arxiv.org/pdf/1811.08996v1.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":"hyperadam-a-learnable-task-adaptive-adam-for","repo_url":"https://github.com/ShipengWang/HyperAdam","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"hyperadam-a-learnable-task-adaptive-adam-for","repo_url":"https://github.com/ShipengWang/HyperAdam-Tensorflow","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"stochastic-optimization","task_name":"Stochastic Optimization"}],"methods":[{"method_slug":"adam","method_name":"Adam"},{"method_slug":"lstm","method_name":"LSTM"},{"method_slug":"sgd","method_name":"SGD"},{"method_slug":"sigmoid-activation","method_name":"Sigmoid Activation"},{"method_slug":"tanh-activation","method_name":"Tanh Activation"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1811.08996","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}