{"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/improving-pairwise-ranking-for-multi-label","title":"Improving Pairwise Ranking for Multi-label Image Classification","arxiv_id":"1704.03135","date":"2017-04-11","proceeding":"CVPR 2017 7","authors":["Yuncheng Li","Yale Song","Jiebo Luo"],"abstract":"Learning to rank has recently emerged as an attractive technique to train\ndeep convolutional neural networks for various computer vision tasks. Pairwise\nranking, in particular, has been successful in multi-label image\nclassification, achieving state-of-the-art results on various benchmarks.\nHowever, most existing approaches use the hinge loss to train their models,\nwhich is non-smooth and thus is difficult to optimize especially with deep\nnetworks. Furthermore, they employ simple heuristics, such as top-k or\nthresholding, to determine which labels to include in the output from a ranked\nlist of labels, which limits their use in the real-world setting. In this work,\nwe propose two techniques to improve pairwise ranking based multi-label image\nclassification: (1) we propose a novel loss function for pairwise ranking,\nwhich is smooth everywhere and thus is easier to optimize; and (2) we\nincorporate a label decision module into the model, estimating the optimal\nconfidence thresholds for each visual concept. We provide theoretical analyses\nof our loss function in the Bayes consistency and risk minimization framework,\nand show its benefit over existing pairwise ranking formulations. We\ndemonstrate the effectiveness of our approach on three large-scale datasets,\nVOC2007, NUS-WIDE and MS-COCO, achieving the best reported results in the\nliterature.","url_abs":"http://arxiv.org/abs/1704.03135v3","url_pdf":"http://arxiv.org/pdf/1704.03135v3.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":"improving-pairwise-ranking-for-multi-label","repo_url":"https://bitbucket.org/raingo-ur/mll-tf","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"improving-pairwise-ranking-for-multi-label","repo_url":"https://github.com/ex4sperans/freesound-classification","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}},{"paper_slug":"improving-pairwise-ranking-for-multi-label","repo_url":"https://github.com/poteminr/agrocode2021","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}},{"paper_slug":"improving-pairwise-ranking-for-multi-label","repo_url":"https://github.com/OFRIN/Tensorflow_Improving_Pairwise_Ranking_for_Multi-label_Image_Classification","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"learning-to-rank","task_name":"Learning-To-Rank"},{"task_slug":"multi-label-image-classification","task_name":"Multi-Label Image Classification"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1704.03135","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}