{"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/self-attentive-residual-decoder-for-neural","title":"Self-Attentive Residual Decoder for Neural Machine Translation","arxiv_id":"1709.04849","date":"2017-09-14","proceeding":"NAACL 2018 6","authors":["Lesly Miculicich Werlen","Nikolaos Pappas","Dhananjay Ram","Andrei Popescu-Belis"],"abstract":"Neural sequence-to-sequence networks with attention have achieved remarkable\nperformance for machine translation. One of the reasons for their effectiveness\nis their ability to capture relevant source-side contextual information at each\ntime-step prediction through an attention mechanism. However, the target-side\ncontext is solely based on the sequence model which, in practice, is prone to a\nrecency bias and lacks the ability to capture effectively non-sequential\ndependencies among words. To address this limitation, we propose a\ntarget-side-attentive residual recurrent network for decoding, where attention\nover previous words contributes directly to the prediction of the next word.\nThe residual learning facilitates the flow of information from the distant past\nand is able to emphasize any of the previously translated words, hence it gains\naccess to a wider context. The proposed model outperforms a neural MT baseline\nas well as a memory and self-attention network on three language pairs. The\nanalysis of the attention learned by the decoder confirms that it emphasizes a\nwider context, and that it captures syntactic-like structures.","url_abs":"http://arxiv.org/abs/1709.04849v5","url_pdf":"http://arxiv.org/pdf/1709.04849v5.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":"self-attentive-residual-decoder-for-neural","repo_url":"https://github.com/idiap/Attentive_Residual_Connections_NMT","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"machine-translation","task_name":"Machine Translation"},{"task_slug":"translation","task_name":"Translation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}