{"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-convolutional-neural-networks-with-merge","title":"Deep Convolutional Neural Networks with Merge-and-Run Mappings","arxiv_id":"1611.07718","date":"2016-11-23","proceeding":null,"authors":["Liming Zhao","Jingdong Wang","Xi Li","Zhuowen Tu","Wen-Jun Zeng"],"abstract":"A deep residual network, built by stacking a sequence of residual blocks, is\neasy to train, because identity mappings skip residual branches and thus\nimprove information flow. To further reduce the training difficulty, we present\na simple network architecture, deep merge-and-run neural networks. The novelty\nlies in a modularized building block, merge-and-run block, which assembles\nresidual branches in parallel through a merge-and-run mapping: Average the\ninputs of these residual branches (Merge), and add the average to the output of\neach residual branch as the input of the subsequent residual branch (Run),\nrespectively. We show that the merge-and-run mapping is a linear idempotent\nfunction in which the transformation matrix is idempotent, and thus improves\ninformation flow, making training easy. In comparison to residual networks, our\nnetworks enjoy compelling advantages: they contain much shorter paths, and the\nwidth, i.e., the number of channels, is increased. We evaluate the performance\non the standard recognition tasks. Our approach demonstrates consistent\nimprovements over ResNets with the comparable setup, and achieves competitive\nresults (e.g., $3.57\\%$ testing error on CIFAR-$10$, $19.00\\%$ on CIFAR-$100$,\n$1.51\\%$ on SVHN).","url_abs":"http://arxiv.org/abs/1611.07718v2","url_pdf":"http://arxiv.org/pdf/1611.07718v2.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-convolutional-neural-networks-with-merge","repo_url":"https://github.com/brycexu/MR-Residual-Net","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"deep-convolutional-neural-networks-with-merge","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":"deep-convolutional-neural-networks-with-merge","repo_url":"https://github.com/lab-midas/med_segmentation","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"deep-convolutional-neural-networks-with-merge","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":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1611.07718","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}