{"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/rebnet-residual-binarized-neural-network","title":"ReBNet: Residual Binarized Neural Network","arxiv_id":"1711.01243","date":"2017-11-03","proceeding":null,"authors":["Mohammad Ghasemzadeh","Mohammad Samragh","Farinaz Koushanfar"],"abstract":"This paper proposes ReBNet, an end-to-end framework for training\nreconfigurable binary neural networks on software and developing efficient\naccelerators for execution on FPGA. Binary neural networks offer an intriguing\nopportunity for deploying large-scale deep learning models on\nresource-constrained devices. Binarization reduces the memory footprint and\nreplaces the power-hungry matrix-multiplication with light-weight XnorPopcount\noperations. However, binary networks suffer from a degraded accuracy compared\nto their fixed-point counterparts. We show that the state-of-the-art methods\nfor optimizing binary networks accuracy, significantly increase the\nimplementation cost and complexity. To compensate for the degraded accuracy\nwhile adhering to the simplicity of binary networks, we devise the first\nreconfigurable scheme that can adjust the classification accuracy based on the\napplication. Our proposition improves the classification accuracy by\nrepresenting features with multiple levels of residual binarization. Unlike\nprevious methods, our approach does not exacerbate the area cost of the\nhardware accelerator. Instead, it provides a tradeoff between throughput and\naccuracy while the area overhead of multi-level binarization is negligible.","url_abs":"http://arxiv.org/abs/1711.01243v3","url_pdf":"http://arxiv.org/pdf/1711.01243v3.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":"rebnet-residual-binarized-neural-network","repo_url":"https://github.com/mohaghasemzadeh/ReBNet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"mxnet","reach":null}],"tasks":[{"task_slug":"binarization","task_name":"Binarization"},{"task_slug":"classification","task_name":"General Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}