{"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/variational-memory-encoder-decoder","title":"Variational Memory Encoder-Decoder","arxiv_id":"1807.09950","date":"2018-07-26","proceeding":"NeurIPS 2018 12","authors":["Hung Le","Truyen Tran","Thin Nguyen","Svetha Venkatesh"],"abstract":"Introducing variability while maintaining coherence is a core task in\nlearning to generate utterances in conversation. Standard neural\nencoder-decoder models and their extensions using conditional variational\nautoencoder often result in either trivial or digressive responses. To overcome\nthis, we explore a novel approach that injects variability into neural\nencoder-decoder via the use of external memory as a mixture model, namely\nVariational Memory Encoder-Decoder (VMED). By associating each memory read with\na mode in the latent mixture distribution at each timestep, our model can\ncapture the variability observed in sequential data such as natural\nconversations. We empirically compare the proposed model against other recent\napproaches on various conversational datasets. The results show that VMED\nconsistently achieves significant improvement over others in both metric-based\nand qualitative evaluations.","url_abs":"http://arxiv.org/abs/1807.09950v2","url_pdf":"http://arxiv.org/pdf/1807.09950v2.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":"variational-memory-encoder-decoder","repo_url":"https://github.com/thaihungle/VMED","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}