{"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/low-shot-learning-with-imprinted-weights","title":"Low-Shot Learning with Imprinted Weights","arxiv_id":"1712.07136","date":"2017-12-19","proceeding":"CVPR 2018 6","authors":["Hang Qi","Matthew Brown","David G. Lowe"],"abstract":"Human vision is able to immediately recognize novel visual categories after\nseeing just one or a few training examples. We describe how to add a similar\ncapability to ConvNet classifiers by directly setting the final layer weights\nfrom novel training examples during low-shot learning. We call this process\nweight imprinting as it directly sets weights for a new category based on an\nappropriately scaled copy of the embedding layer activations for that training\nexample. The imprinting process provides a valuable complement to training with\nstochastic gradient descent, as it provides immediate good classification\nperformance and an initialization for any further fine-tuning in the future. We\nshow how this imprinting process is related to proxy-based embeddings. However,\nit differs in that only a single imprinted weight vector is learned for each\nnovel category, rather than relying on a nearest-neighbor distance to training\ninstances as typically used with embedding methods. Our experiments show that\nusing averaging of imprinted weights provides better generalization than using\nnearest-neighbor instance embeddings.","url_abs":"http://arxiv.org/abs/1712.07136v2","url_pdf":"http://arxiv.org/pdf/1712.07136v2.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":"low-shot-learning-with-imprinted-weights","repo_url":"https://github.com/Kao1126/edgetpu-TransferLearning","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}