{"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/neural-map-structured-memory-for-deep","title":"Neural Map: Structured Memory for Deep Reinforcement Learning","arxiv_id":"1702.08360","date":"2017-02-27","proceeding":"ICLR 2018 1","authors":["Emilio Parisotto","Ruslan Salakhutdinov"],"abstract":"A critical component to enabling intelligent reasoning in partially\nobservable environments is memory. Despite this importance, Deep Reinforcement\nLearning (DRL) agents have so far used relatively simple memory architectures,\nwith the main methods to overcome partial observability being either a temporal\nconvolution over the past k frames or an LSTM layer. More recent work (Oh et\nal., 2016) has went beyond these architectures by using memory networks which\ncan allow more sophisticated addressing schemes over the past k frames. But\neven these architectures are unsatisfactory due to the reason that they are\nlimited to only remembering information from the last k frames. In this paper,\nwe develop a memory system with an adaptable write operator that is customized\nto the sorts of 3D environments that DRL agents typically interact with. This\narchitecture, called the Neural Map, uses a spatially structured 2D memory\nimage to learn to store arbitrary information about the environment over long\ntime lags. We demonstrate empirically that the Neural Map surpasses previous\nDRL memories on a set of challenging 2D and 3D maze environments and show that\nit is capable of generalizing to environments that were not seen during\ntraining.","url_abs":"http://arxiv.org/abs/1702.08360v1","url_pdf":"http://arxiv.org/pdf/1702.08360v1.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":"neural-map-structured-memory-for-deep","repo_url":"https://github.com/zuoxingdong/VIN_PyTorch_Visdom","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"deep-reinforcement-learning","task_name":"Deep Reinforcement Learning"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"},{"task_slug":"reinforcement-learning-2","task_name":"reinforcement-learning"}],"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=1702.08360","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}