{"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/linking-image-and-text-with-2-way-nets","title":"Linking Image and Text with 2-Way Nets","arxiv_id":"1608.07973","date":"2016-08-29","proceeding":"CVPR 2017 7","authors":["Aviv Eisenschtat","Lior Wolf"],"abstract":"Linking two data sources is a basic building block in numerous computer\nvision problems. Canonical Correlation Analysis (CCA) achieves this by\nutilizing a linear optimizer in order to maximize the correlation between the\ntwo views. Recent work makes use of non-linear models, including deep learning\ntechniques, that optimize the CCA loss in some feature space. In this paper, we\nintroduce a novel, bi-directional neural network architecture for the task of\nmatching vectors from two data sources. Our approach employs two tied neural\nnetwork channels that project the two views into a common, maximally correlated\nspace using the Euclidean loss. We show a direct link between the\ncorrelation-based loss and Euclidean loss, enabling the use of Euclidean loss\nfor correlation maximization. To overcome common Euclidean regression\noptimization problems, we modify well-known techniques to our problem,\nincluding batch normalization and dropout. We show state of the art results on\na number of computer vision matching tasks including MNIST image matching and\nsentence-image matching on the Flickr8k, Flickr30k and COCO datasets.","url_abs":"http://arxiv.org/abs/1608.07973v3","url_pdf":"http://arxiv.org/pdf/1608.07973v3.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":"linking-image-and-text-with-2-way-nets","repo_url":"https://github.com/aviveise/2WayNet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"sentence","task_name":"Sentence"}],"methods":[{"method_slug":"batch-normalization","method_name":"Batch Normalization"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-retrieval-on-flickr30k-1k-test","task":"Image Retrieval","dataset":"Flickr30K 1K test","model":"2WayNet (VGG)","rank_in_archive_order":13,"of":18,"metrics":{"R@1":"36.0"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1608.07973","atlas_url":"https://app.syntology.ai/?focus=1608.07973","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}