{"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/learning-long-term-dependencies-for-action","title":"Learning long-term dependencies for action recognition with a biologically-inspired deep network","arxiv_id":"1611.05216","date":"2016-11-16","proceeding":"ICCV 2017 10","authors":["Yemin Shi","Yonghong Tian","Yao-Wei Wang","Tiejun Huang"],"abstract":"Despite a lot of research efforts devoted in recent years, how to efficiently\nlearn long-term dependencies from sequences still remains a pretty challenging\ntask. As one of the key models for sequence learning, recurrent neural network\n(RNN) and its variants such as long short term memory (LSTM) and gated\nrecurrent unit (GRU) are still not powerful enough in practice. One possible\nreason is that they have only feedforward connections, which is different from\nthe biological neural system that is typically composed of both feedforward and\nfeedback connections. To address this problem, this paper proposes a\nbiologically-inspired deep network, called shuttleNet\\footnote{Our code is\navailable at \\url{https://github.com/shiyemin/shuttlenet}}. Technologically,\nthe shuttleNet consists of several processors, each of which is a GRU while\nassociated with multiple groups of cells and states. Unlike traditional RNNs,\nall processors inside shuttleNet are loop connected to mimic the brain's\nfeedforward and feedback connections, in which they are shared across multiple\npathways in the loop connection. Attention mechanism is then employed to select\nthe best information flow pathway. Extensive experiments conducted on two\nbenchmark datasets (i.e UCF101 and HMDB51) show that we can beat\nstate-of-the-art methods by simply embedding shuttleNet into a CNN-RNN\nframework.","url_abs":"http://arxiv.org/abs/1611.05216v3","url_pdf":"http://arxiv.org/pdf/1611.05216v3.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":"learning-long-term-dependencies-for-action","repo_url":"https://github.com/shiyemin/shuttlenet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"action-recognition-in-videos","task_name":"Action Recognition"},{"task_slug":"action-recognition","task_name":"Temporal Action Localization"}],"methods":[{"method_slug":"gru","method_name":"GRU"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1611.05216","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}