{"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/hierarchically-structured-reinforcement","title":"Hierarchically Structured Reinforcement Learning for Topically Coherent Visual Story Generation","arxiv_id":"1805.08191","date":"2018-05-21","proceeding":null,"authors":["Qiuyuan Huang","Zhe Gan","Asli Celikyilmaz","Dapeng Wu","Jian-Feng Wang","Xiaodong He"],"abstract":"We propose a hierarchically structured reinforcement learning approach to\naddress the challenges of planning for generating coherent multi-sentence\nstories for the visual storytelling task. Within our framework, the task of\ngenerating a story given a sequence of images is divided across a two-level\nhierarchical decoder. The high-level decoder constructs a plan by generating a\nsemantic concept (i.e., topic) for each image in sequence. The low-level\ndecoder generates a sentence for each image using a semantic compositional\nnetwork, which effectively grounds the sentence generation conditioned on the\ntopic. The two decoders are jointly trained end-to-end using reinforcement\nlearning. We evaluate our model on the visual storytelling (VIST) dataset.\nEmpirical results from both automatic and human evaluations demonstrate that\nthe proposed hierarchically structured reinforced training achieves\nsignificantly better performance compared to a strong flat deep reinforcement\nlearning baseline.","url_abs":"http://arxiv.org/abs/1805.08191v3","url_pdf":"http://arxiv.org/pdf/1805.08191v3.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":[],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"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":"sentence","task_name":"Sentence"},{"task_slug":"story-generation","task_name":"Story Generation"},{"task_slug":"visual-storytelling","task_name":"Visual Storytelling"},{"task_slug":"reinforcement-learning-2","task_name":"reinforcement-learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/visual-storytelling-on-vist","task":"Visual Storytelling","dataset":"VIST","model":"HSRL w/ Joint Training","rank_in_archive_order":24,"of":33,"metrics":{"BLEU-4":"12.32","CIDEr":"10.71","METEOR":"35.23","ROUGE-L":"30.84","SPICE":"12.97"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1805.08191","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}