{"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/deep-competitive-pathway-networks","title":"Deep Competitive Pathway Networks","arxiv_id":"1709.10282","date":"2017-09-29","proceeding":null,"authors":["Jia-Ren Chang","Yong-Sheng Chen"],"abstract":"In the design of deep neural architectures, recent studies have demonstrated\nthe benefits of grouping subnetworks into a larger network. For examples, the\nInception architecture integrates multi-scale subnetworks and the residual\nnetwork can be regarded that a residual unit combines a residual subnetwork\nwith an identity shortcut. In this work, we embrace this observation and\npropose the Competitive Pathway Network (CoPaNet). The CoPaNet comprises a\nstack of competitive pathway units and each unit contains multiple parallel\nresidual-type subnetworks followed by a max operation for feature competition.\nThis mechanism enhances the model capability by learning a variety of features\nin subnetworks. The proposed strategy explicitly shows that the features\npropagate through pathways in various routing patterns, which is referred to as\npathway encoding of category information. Moreover, the cross-block shortcut\ncan be added to the CoPaNet to encourage feature reuse. We evaluated the\nproposed CoPaNet on four object recognition benchmarks: CIFAR-10, CIFAR-100,\nSVHN, and ImageNet. CoPaNet obtained the state-of-the-art or comparable results\nusing similar amounts of parameters. The code of CoPaNet is available at:\nhttps://github.com/JiaRenChang/CoPaNet.","url_abs":"http://arxiv.org/abs/1709.10282v1","url_pdf":"http://arxiv.org/pdf/1709.10282v1.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":"deep-competitive-pathway-networks","repo_url":"https://github.com/JiaRenChang/CoPaNet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"torch","reach":null}],"tasks":[{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"object-recognition","task_name":"Object Recognition"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-classification-on-cifar-10","task":"Image Classification","dataset":"CIFAR-10","model":"CoPaNet-R-164","rank_in_archive_order":105,"of":265,"metrics":{"Percentage correct":"96.62"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-cifar-100","task":"Image Classification","dataset":"CIFAR-100","model":"CoPaNet-R-164","rank_in_archive_order":123,"of":211,"metrics":{"Percentage correct":"81.10"},"uses_additional_data":true},{"leaderboard":"/sota/image-classification-on-svhn","task":"Image Classification","dataset":"SVHN","model":"CoPaNet-R-164","rank_in_archive_order":13,"of":62,"metrics":{"Percentage error":"1.58"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}