{"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/factorized-bilinear-models-for-image","title":"Factorized Bilinear Models for Image Recognition","arxiv_id":"1611.05709","date":"2016-11-17","proceeding":"ICCV 2017 10","authors":["Yanghao Li","Naiyan Wang","Jiaying Liu","Xiaodi Hou"],"abstract":"Although Deep Convolutional Neural Networks (CNNs) have liberated their power\nin various computer vision tasks, the most important components of CNN,\nconvolutional layers and fully connected layers, are still limited to linear\ntransformations. In this paper, we propose a novel Factorized Bilinear (FB)\nlayer to model the pairwise feature interactions by considering the quadratic\nterms in the transformations. Compared with existing methods that tried to\nincorporate complex non-linearity structures into CNNs, the factorized\nparameterization makes our FB layer only require a linear increase of\nparameters and affordable computational cost. To further reduce the risk of\noverfitting of the FB layer, a specific remedy called DropFactor is devised\nduring the training process. We also analyze the connection between FB layer\nand some existing models, and show FB layer is a generalization to them.\nFinally, we validate the effectiveness of FB layer on several widely adopted\ndatasets including CIFAR-10, CIFAR-100 and ImageNet, and demonstrate superior\nresults compared with various state-of-the-art deep models.","url_abs":"http://arxiv.org/abs/1611.05709v2","url_pdf":"http://arxiv.org/pdf/1611.05709v2.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":"factorized-bilinear-models-for-image","repo_url":"https://github.com/lyttonhao/Factorized-Bilinear-Network","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"mxnet","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1611.05709","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}