{"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/simplifying-neural-machine-translation-with","title":"Simplifying Neural Machine Translation with Addition-Subtraction Twin-Gated Recurrent Networks","arxiv_id":"1810.12546","date":"2018-10-30","proceeding":"EMNLP 2018 10","authors":["Biao Zhang","Deyi Xiong","Jinsong Su","Qian Lin","Huiji Zhang"],"abstract":"In this paper, we propose an additionsubtraction twin-gated recurrent network\n(ATR) to simplify neural machine translation. The recurrent units of ATR are\nheavily simplified to have the smallest number of weight matrices among units\nof all existing gated RNNs. With the simple addition and subtraction operation,\nwe introduce a twin-gated mechanism to build input and forget gates which are\nhighly correlated. Despite this simplification, the essential non-linearities\nand capability of modeling long-distance dependencies are preserved.\nAdditionally, the proposed ATR is more transparent than LSTM/GRU due to the\nsimplification. Forward self-attention can be easily established in ATR, which\nmakes the proposed network interpretable. Experiments on WMT14 translation\ntasks demonstrate that ATR-based neural machine translation can yield\ncompetitive performance on English- German and English-French language pairs in\nterms of both translation quality and speed. Further experiments on NIST\nChinese-English translation, natural language inference and Chinese word\nsegmentation verify the generality and applicability of ATR on different\nnatural language processing tasks.","url_abs":"http://arxiv.org/abs/1810.12546v1","url_pdf":"http://arxiv.org/pdf/1810.12546v1.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":"simplifying-neural-machine-translation-with","repo_url":"https://github.com/bzhangGo/ATR","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null},{"paper_slug":"simplifying-neural-machine-translation-with","repo_url":"https://github.com/bzhangGo/zero","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null},{"paper_slug":"simplifying-neural-machine-translation-with","repo_url":"https://github.com/bzhangGo/lrn","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[{"task_slug":"chinese-word-segmentation","task_name":"Chinese Word Segmentation"},{"task_slug":"machine-translation","task_name":"Machine Translation"},{"task_slug":"natural-language-inference","task_name":"Natural Language Inference"},{"task_slug":"translation","task_name":"Translation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}