{"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/supervised-hashing-using-graph-cuts-and","title":"Supervised Hashing Using Graph Cuts and Boosted Decision Trees","arxiv_id":"1408.5574","date":"2014-08-24","proceeding":null,"authors":["Guosheng Lin","Chunhua Shen","Anton Van Den Hengel"],"abstract":"Embedding image features into a binary Hamming space can improve both the\nspeed and accuracy of large-scale query-by-example image retrieval systems.\nSupervised hashing aims to map the original features to compact binary codes in\na manner which preserves the label-based similarities of the original data.\nMost existing approaches apply a single form of hash function, and an\noptimization process which is typically deeply coupled to this specific form.\nThis tight coupling restricts the flexibility of those methods, and can result\nin complex optimization problems that are difficult to solve. In this work we\nproffer a flexible yet simple framework that is able to accommodate different\ntypes of loss functions and hash functions. The proposed framework allows a\nnumber of existing approaches to hashing to be placed in context, and\nsimplifies the development of new problem-specific hashing methods. Our\nframework decomposes the into two steps: binary code (hash bits) learning, and\nhash function learning. The first step can typically be formulated as a binary\nquadratic problem, and the second step can be accomplished by training standard\nbinary classifiers. For solving large-scale binary code inference, we show how\nto ensure that the binary quadratic problems are submodular such that an\nefficient graph cut approach can be used. To achieve efficiency as well as\nefficacy on large-scale high-dimensional data, we propose to use boosted\ndecision trees as the hash functions, which are nonlinear, highly descriptive,\nand very fast to train and evaluate. Experiments demonstrate that our proposed\nmethod significantly outperforms most state-of-the-art methods, especially on\nhigh-dimensional data.","url_abs":"http://arxiv.org/abs/1408.5574v2","url_pdf":"http://arxiv.org/pdf/1408.5574v2.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":"supervised-hashing-using-graph-cuts-and","repo_url":"https://bitbucket.org/chhshen/fasthash","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"descriptive","task_name":"Descriptive"},{"task_slug":"image-retrieval","task_name":"Image Retrieval"},{"task_slug":"retrieval","task_name":"Retrieval"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}