{"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/on-the-effectiveness-of-task-granularity-for","title":"On the effectiveness of task granularity for transfer learning","arxiv_id":"1804.09235","date":"2018-04-24","proceeding":null,"authors":["Farzaneh Mahdisoltani","Guillaume Berger","Waseem Gharbieh","David Fleet","Roland Memisevic"],"abstract":"We describe a DNN for video classification and captioning, trained\nend-to-end, with shared features, to solve tasks at different levels of\ngranularity, exploring the link between granularity in a source task and the\nquality of learned features for transfer learning. For solving the new task\ndomain in transfer learning, we freeze the trained encoder and fine-tune a\nneural net on the target domain. We train on the Something-Something dataset\nwith over 220, 000 videos, and multiple levels of target granularity, including\n50 action groups, 174 fine-grained action categories and captions.\nClassification and captioning with Something-Something are challenging because\nof the subtle differences between actions, applied to thousands of different\nobject classes, and the diversity of captions penned by crowd actors. Our model\nperforms better than existing classification baselines for SomethingSomething,\nwith impressive fine-grained results. And it yields a strong baseline on the\nnew Something-Something captioning task. Experiments reveal that training with\nmore fine-grained tasks tends to produce better features for transfer learning.","url_abs":"http://arxiv.org/abs/1804.09235v2","url_pdf":"http://arxiv.org/pdf/1804.09235v2.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":"on-the-effectiveness-of-task-granularity-for","repo_url":"https://github.com/jssprz/video_captioning_datasets","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"diversity","task_name":"Diversity"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"transfer-learning","task_name":"Transfer Learning"},{"task_slug":"video-classification","task_name":"Video Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1804.09235","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}