{"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/deeply-fused-nets","title":"Deeply-Fused Nets","arxiv_id":"1605.07716","date":"2016-05-25","proceeding":null,"authors":["Jingdong Wang","Zhen Wei","Ting Zhang","Wen-Jun Zeng"],"abstract":"In this paper, we present a novel deep learning approach, deeply-fused nets.\nThe central idea of our approach is deep fusion, i.e., combine the intermediate\nrepresentations of base networks, where the fused output serves as the input of\nthe remaining part of each base network, and perform such combinations deeply\nover several intermediate representations. The resulting deeply fused net\nenjoys several benefits. First, it is able to learn multi-scale representations\nas it enjoys the benefits of more base networks, which could form the same\nfused network, other than the initial group of base networks. Second, in our\nsuggested fused net formed by one deep and one shallow base networks, the flows\nof the information from the earlier intermediate layer of the deep base network\nto the output and from the input to the later intermediate layer of the deep\nbase network are both improved. Last, the deep and shallow base networks are\njointly learnt and can benefit from each other. More interestingly, the\nessential depth of a fused net composed from a deep base network and a shallow\nbase network is reduced because the fused net could be composed from a less\ndeep base network, and thus training the fused net is less difficult than\ntraining the initial deep base network. Empirical results demonstrate that our\napproach achieves superior performance over two closely-related methods, ResNet\nand Highway, and competitive performance compared to the state-of-the-arts.","url_abs":"http://arxiv.org/abs/1605.07716v1","url_pdf":"http://arxiv.org/pdf/1605.07716v1.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":"deeply-fused-nets","repo_url":"https://github.com/homles11/IGCV3","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"deeply-fused-nets","repo_url":"https://github.com/zlmzju/fusenet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1605.07716","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}