{"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/statistically-motivated-second-order-pooling-1","title":"Statistically Motivated Second Order Pooling","arxiv_id":"1801.07492","date":"2018-01-23","proceeding":null,"authors":["Kaicheng Yu","Mathieu Salzmann"],"abstract":"Second-order pooling, a.k.a.~bilinear pooling, has proven effective for deep\nlearning based visual recognition. However, the resulting second-order networks\nyield a final representation that is orders of magnitude larger than that of\nstandard, first-order ones, making them memory-intensive and cumbersome to\ndeploy. Here, we introduce a general, parametric compression strategy that can\nproduce more compact representations than existing compression techniques, yet\noutperform both compressed and uncompressed second-order models. Our approach\nis motivated by a statistical analysis of the network's activations, relying on\noperations that lead to a Gaussian-distributed final representation, as\ninherently used by first-order deep networks. As evidenced by our experiments,\nthis lets us outperform the state-of-the-art first-order and second-order\nmodels on several benchmark recognition datasets.","url_abs":"http://arxiv.org/abs/1801.07492v3","url_pdf":"http://arxiv.org/pdf/1801.07492v3.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":"statistically-motivated-second-order-pooling-1","repo_url":"https://github.com/kcyu2014/smsop","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1801.07492","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}