{"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/sparsely-aggregated-convolutional-networks","title":"Sparsely Aggregated Convolutional Networks","arxiv_id":"1801.05895","date":"2018-01-18","proceeding":"ECCV 2018 9","authors":["Ligeng Zhu","Ruizhi Deng","Michael Maire","Zhiwei Deng","Greg Mori","Ping Tan"],"abstract":"We explore a key architectural aspect of deep convolutional neural networks:\nthe pattern of internal skip connections used to aggregate outputs of earlier\nlayers for consumption by deeper layers. Such aggregation is critical to\nfacilitate training of very deep networks in an end-to-end manner. This is a\nprimary reason for the widespread adoption of residual networks, which\naggregate outputs via cumulative summation. While subsequent works investigate\nalternative aggregation operations (e.g. concatenation), we focus on an\northogonal question: which outputs to aggregate at a particular point in the\nnetwork. We propose a new internal connection structure which aggregates only a\nsparse set of previous outputs at any given depth. Our experiments demonstrate\nthis simple design change offers superior performance with fewer parameters and\nlower computational requirements. Moreover, we show that sparse aggregation\nallows networks to scale more robustly to 1000+ layers, thereby opening future\navenues for training long-running visual processes.","url_abs":"http://arxiv.org/abs/1801.05895v3","url_pdf":"http://arxiv.org/pdf/1801.05895v3.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":"sparsely-aggregated-convolutional-networks","repo_url":"https://github.com/Lyken17/SparseNet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"sparsely-aggregated-convolutional-networks","repo_url":"https://github.com/osmr/imgclsmob","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"mxnet","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}