{"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/multi-level-semantic-feature-augmentation-for","title":"Multi-level Semantic Feature Augmentation for One-shot Learning","arxiv_id":"1804.05298","date":"2018-04-15","proceeding":null,"authors":["Zitian Chen","Yanwei Fu","yinda zhang","Yu-Gang Jiang","xiangyang xue","Leonid Sigal"],"abstract":"The ability to quickly recognize and learn new visual concepts from limited\nsamples enables humans to swiftly adapt to new environments. This ability is\nenabled by semantic associations of novel concepts with those that have already\nbeen learned and stored in memory. Computers can start to ascertain similar\nabilities by utilizing a semantic concept space. A concept space is a\nhigh-dimensional semantic space in which similar abstract concepts appear close\nand dissimilar ones far apart. In this paper, we propose a novel approach to\none-shot learning that builds on this idea. Our approach learns to map a novel\nsample instance to a concept, relates that concept to the existing ones in the\nconcept space and generates new instances, by interpolating among the concepts,\nto help learning. Instead of synthesizing new image instance, we propose to\ndirectly synthesize instance features by leveraging semantics using a novel\nauto-encoder network we call dual TriNet. The encoder part of the TriNet learns\nto map multi-layer visual features of deep CNNs, that is, multi-level concepts,\nto a semantic vector. In semantic space, we search for related concepts, which\nare then projected back into the image feature spaces by the decoder portion of\nthe TriNet. Two strategies in the semantic space are explored. Notably, this\nseemingly simple strategy results in complex augmented feature distributions in\nthe image feature space, leading to substantially better performance.","url_abs":"http://arxiv.org/abs/1804.05298v4","url_pdf":"http://arxiv.org/pdf/1804.05298v4.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":"multi-level-semantic-feature-augmentation-for","repo_url":"https://github.com/tankche1/Semantic-Feature-Augmentation-in-Few-shot-Learning","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"novel-concepts","task_name":"Novel Concepts"},{"task_slug":"one-shot-learning","task_name":"One-Shot Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1804.05298","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}