{"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/learning-without-memorizing","title":"Learning without Memorizing","arxiv_id":"1811.08051","date":"2018-11-20","proceeding":"CVPR 2019 6","authors":["Prithviraj Dhar","Rajat Vikram Singh","Kuan-Chuan Peng","Ziyan Wu","Rama Chellappa"],"abstract":"Incremental learning (IL) is an important task aimed at increasing the\ncapability of a trained model, in terms of the number of classes recognizable\nby the model. The key problem in this task is the requirement of storing data\n(e.g. images) associated with existing classes, while teaching the classifier\nto learn new classes. However, this is impractical as it increases the memory\nrequirement at every incremental step, which makes it impossible to implement\nIL algorithms on edge devices with limited memory. Hence, we propose a novel\napproach, called `Learning without Memorizing (LwM)', to preserve the\ninformation about existing (base) classes, without storing any of their data,\nwhile making the classifier progressively learn the new classes. In LwM, we\npresent an information preserving penalty: Attention Distillation Loss\n($L_{AD}$), and demonstrate that penalizing the changes in classifiers'\nattention maps helps to retain information of the base classes, as new classes\nare added. We show that adding $L_{AD}$ to the distillation loss which is an\nexisting information preserving loss consistently outperforms the\nstate-of-the-art performance in the iILSVRC-small and iCIFAR-100 datasets in\nterms of the overall accuracy of base and incrementally learned classes.","url_abs":"http://arxiv.org/abs/1811.08051v2","url_pdf":"http://arxiv.org/pdf/1811.08051v2.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":"learning-without-memorizing","repo_url":"https://github.com/mmasana/FACIL","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"incremental-learning","task_name":"Incremental Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1811.08051","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}