{"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/dual-memory-neural-computer-for-asynchronous","title":"Dual Memory Neural Computer for Asynchronous Two-view Sequential Learning","arxiv_id":"1802.00662","date":"2018-02-02","proceeding":null,"authors":["Hung Le","Truyen Tran","Svetha Venkatesh"],"abstract":"One of the core tasks in multi-view learning is to capture relations among\nviews. For sequential data, the relations not only span across views, but also\nextend throughout the view length to form long-term intra-view and inter-view\ninteractions. In this paper, we present a new memory augmented neural network\nmodel that aims to model these complex interactions between two asynchronous\nsequential views. Our model uses two encoders for reading from and writing to\ntwo external memories for encoding input views. The intra-view interactions and\nthe long-term dependencies are captured by the use of memories during this\nencoding process. There are two modes of memory accessing in our system:\nlate-fusion and early-fusion, corresponding to late and early inter-view\ninteractions. In the late-fusion mode, the two memories are separated,\ncontaining only view-specific contents. In the early-fusion mode, the two\nmemories share the same addressing space, allowing cross-memory accessing. In\nboth cases, the knowledge from the memories will be combined by a decoder to\nmake predictions over the output space. The resulting dual memory neural\ncomputer is demonstrated on a comprehensive set of experiments, including a\nsynthetic task of summing two sequences and the tasks of drug prescription and\ndisease progression in healthcare. The results demonstrate competitive\nperformance over both traditional algorithms and deep learning methods designed\nfor multi-view problems.","url_abs":"http://arxiv.org/abs/1802.00662v2","url_pdf":"http://arxiv.org/pdf/1802.00662v2.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":"dual-memory-neural-computer-for-asynchronous","repo_url":"https://github.com/thaihungle/DMNC","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"multi-view-learning","task_name":"MULTI-VIEW LEARNING"},{"task_slug":"two","task_name":"Vocal Bursts Valence Prediction"}],"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}