{"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/unifying-visual-semantic-embeddings-with","title":"Unifying Visual-Semantic Embeddings with Multimodal Neural Language Models","arxiv_id":"1411.2539","date":"2014-11-10","proceeding":null,"authors":["Ryan Kiros","Ruslan Salakhutdinov","Richard S. Zemel"],"abstract":"Inspired by recent advances in multimodal learning and machine translation,\nwe introduce an encoder-decoder pipeline that learns (a): a multimodal joint\nembedding space with images and text and (b): a novel language model for\ndecoding distributed representations from our space. Our pipeline effectively\nunifies joint image-text embedding models with multimodal neural language\nmodels. We introduce the structure-content neural language model that\ndisentangles the structure of a sentence to its content, conditioned on\nrepresentations produced by the encoder. The encoder allows one to rank images\nand sentences while the decoder can generate novel descriptions from scratch.\nUsing LSTM to encode sentences, we match the state-of-the-art performance on\nFlickr8K and Flickr30K without using object detections. We also set new best\nresults when using the 19-layer Oxford convolutional network. Furthermore we\nshow that with linear encoders, the learned embedding space captures multimodal\nregularities in terms of vector space arithmetic e.g. *image of a blue car* -\n\"blue\" + \"red\" is near images of red cars. Sample captions generated for 800\nimages are made available for comparison.","url_abs":"http://arxiv.org/abs/1411.2539v1","url_pdf":"http://arxiv.org/pdf/1411.2539v1.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":"unifying-visual-semantic-embeddings-with","repo_url":"https://github.com/Chloejay/image_caption_app","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"unifying-visual-semantic-embeddings-with","repo_url":"https://github.com/chris4540/DD2430-ds-proj","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"NOASSERTION"}},{"paper_slug":"unifying-visual-semantic-embeddings-with","repo_url":"https://github.com/woozzu/dong_iccv_2017","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"language-modeling","task_name":"Language Modeling"},{"task_slug":"language-modelling","task_name":"Language Modelling"},{"task_slug":"machine-translation","task_name":"Machine Translation"},{"task_slug":"sentence","task_name":"Sentence"},{"task_slug":"translation","task_name":"Translation"}],"methods":[{"method_slug":"lstm","method_name":"LSTM"},{"method_slug":"sigmoid-activation","method_name":"Sigmoid Activation"},{"method_slug":"tanh-activation","method_name":"Tanh Activation"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1411.2539","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}