{"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/supportnet-solving-catastrophic-forgetting-in","title":"SupportNet: solving catastrophic forgetting in class incremental learning with support data","arxiv_id":"1806.02942","date":"2018-06-08","proceeding":null,"authors":["Yu Li","Zhongxiao Li","Lizhong Ding","Yijie Pan","Chao Huang","Yuhui Hu","Wei Chen","Xin Gao"],"abstract":"A plain well-trained deep learning model often does not have the ability to\nlearn new knowledge without forgetting the previously learned knowledge, which\nis known as catastrophic forgetting. Here we propose a novel method,\nSupportNet, to efficiently and effectively solve the catastrophic forgetting\nproblem in the class incremental learning scenario. SupportNet combines the\nstrength of deep learning and support vector machine (SVM), where SVM is used\nto identify the support data from the old data, which are fed to the deep\nlearning model together with the new data for further training so that the\nmodel can review the essential information of the old data when learning the\nnew information. Two powerful consolidation regularizers are applied to\nstabilize the learned representation and ensure the robustness of the learned\nmodel. We validate our method with comprehensive experiments on various tasks,\nwhich show that SupportNet drastically outperforms the state-of-the-art\nincremental learning methods and even reaches similar performance as the deep\nlearning model trained from scratch on both old and new data. Our program is\naccessible at: https://github.com/lykaust15/SupportNet","url_abs":"http://arxiv.org/abs/1806.02942v3","url_pdf":"http://arxiv.org/pdf/1806.02942v3.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":"supportnet-solving-catastrophic-forgetting-in","repo_url":"https://github.com/lykaust15/SupportNet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"class-incremental-learning","task_name":"Class Incremental Learning"},{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":"incremental-learning","task_name":"Incremental Learning"},{"task_slug":"class-incremental-learning-1","task_name":"class-incremental learning"}],"methods":[{"method_slug":"svm","method_name":"SVM"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}