{"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/towards-accurate-binary-convolutional-neural","title":"Towards Accurate Binary Convolutional Neural Network","arxiv_id":"1711.11294","date":"2017-11-30","proceeding":"NeurIPS 2017 12","authors":["Xiaofan Lin","Cong Zhao","Wei Pan"],"abstract":"We introduce a novel scheme to train binary convolutional neural networks\n(CNNs) -- CNNs with weights and activations constrained to {-1,+1} at run-time.\nIt has been known that using binary weights and activations drastically reduce\nmemory size and accesses, and can replace arithmetic operations with more\nefficient bitwise operations, leading to much faster test-time inference and\nlower power consumption. However, previous works on binarizing CNNs usually\nresult in severe prediction accuracy degradation. In this paper, we address\nthis issue with two major innovations: (1) approximating full-precision weights\nwith the linear combination of multiple binary weight bases; (2) employing\nmultiple binary activations to alleviate information loss. The implementation\nof the resulting binary CNN, denoted as ABC-Net, is shown to achieve much\ncloser performance to its full-precision counterpart, and even reach the\ncomparable prediction accuracy on ImageNet and forest trail datasets, given\nadequate binary weight bases and activations.","url_abs":"http://arxiv.org/abs/1711.11294v1","url_pdf":"http://arxiv.org/pdf/1711.11294v1.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":"towards-accurate-binary-convolutional-neural","repo_url":"https://github.com/kazimuth/6.338","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"unanswered"}},{"paper_slug":"towards-accurate-binary-convolutional-neural","repo_url":"https://github.com/layog/accurate-binary-convolution-network","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"unanswered"}}],"tasks":[],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1711.11294","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}