{"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/generalization-tower-network-a-novel-deep","title":"Generalization Tower Network: A Novel Deep Neural Network Architecture for Multi-Task Learning","arxiv_id":"1710.10036","date":"2017-10-27","proceeding":null,"authors":["Yuhang Song","Main Xu","Songyang Zhang","Liangyu Huo"],"abstract":"Deep learning (DL) advances state-of-the-art reinforcement learning (RL), by\nincorporating deep neural networks in learning representations from the input\nto RL. However, the conventional deep neural network architecture is limited in\nlearning representations for multi-task RL (MT-RL), as multiple tasks can refer\nto different kinds of representations. In this paper, we thus propose a novel\ndeep neural network architecture, namely generalization tower network (GTN),\nwhich can achieve MT-RL within a single learned model. Specifically, the\narchitecture of GTN is composed of both horizontal and vertical streams. In our\nGTN architecture, horizontal streams are used to learn representation shared in\nsimilar tasks. In contrast, the vertical streams are introduced to be more\nsuitable for handling diverse tasks, which encodes hierarchical shared\nknowledge of these tasks. The effectiveness of the introduced vertical stream\nis validated by experimental results. Experimental results further verify that\nour GTN architecture is able to advance the state-of-the-art MT-RL, via being\ntested on 51 Atari games.","url_abs":"http://arxiv.org/abs/1710.10036v3","url_pdf":"http://arxiv.org/pdf/1710.10036v3.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":"generalization-tower-network-a-novel-deep","repo_url":"https://github.com/YuhangSong/GTN","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"atari-games","task_name":"Atari Games"},{"task_slug":"multi-task-learning","task_name":"Multi-Task Learning"},{"task_slug":"reinforcement-learning","task_name":"Reinforcement Learning"},{"task_slug":"reinforcement-learning-1","task_name":"Reinforcement Learning (RL)"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}