{"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/curriculum-dropout","title":"Curriculum Dropout","arxiv_id":"1703.06229","date":"2017-03-18","proceeding":"ICCV 2017 10","authors":["Pietro Morerio","Jacopo Cavazza","Riccardo Volpi","Rene Vidal","Vittorio Murino"],"abstract":"Dropout is a very effective way of regularizing neural networks.\nStochastically \"dropping out\" units with a certain probability discourages\nover-specific co-adaptations of feature detectors, preventing overfitting and\nimproving network generalization. Besides, Dropout can be interpreted as an\napproximate model aggregation technique, where an exponential number of smaller\nnetworks are averaged in order to get a more powerful ensemble. In this paper,\nwe show that using a fixed dropout probability during training is a suboptimal\nchoice. We thus propose a time scheduling for the probability of retaining\nneurons in the network. This induces an adaptive regularization scheme that\nsmoothly increases the difficulty of the optimization problem. This idea of\n\"starting easy\" and adaptively increasing the difficulty of the learning\nproblem has its roots in curriculum learning and allows one to train better\nmodels. Indeed, we prove that our optimization strategy implements a very\ngeneral curriculum scheme, by gradually adding noise to both the input and\nintermediate feature representations within the network architecture.\nExperiments on seven image classification datasets and different network\narchitectures show that our method, named Curriculum Dropout, frequently yields\nto better generalization and, at worst, performs just as well as the standard\nDropout method.","url_abs":"http://arxiv.org/abs/1703.06229v2","url_pdf":"http://arxiv.org/pdf/1703.06229v2.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":"curriculum-dropout","repo_url":"https://github.com/pmorerio/curriculum-dropout","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok","spdx":"GPL-3.0"}},{"paper_slug":"curriculum-dropout","repo_url":"https://github.com/balajiselvaraj1601/Step_Dropout","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"scheduling","task_name":"Scheduling"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[{"method_slug":"dropout","method_name":"Dropout"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1703.06229","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}