{"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/loss-aware-binarization-of-deep-networks","title":"Loss-aware Binarization of Deep Networks","arxiv_id":"1611.01600","date":"2016-11-05","proceeding":null,"authors":["Lu Hou","Quanming Yao","James T. Kwok"],"abstract":"Deep neural network models, though very powerful and highly successful, are\ncomputationally expensive in terms of space and time. Recently, there have been\na number of attempts on binarizing the network weights and activations. This\ngreatly reduces the network size, and replaces the underlying multiplications\nto additions or even XNOR bit operations. However, existing binarization\nschemes are based on simple matrix approximation and ignore the effect of\nbinarization on the loss. In this paper, we propose a proximal Newton algorithm\nwith diagonal Hessian approximation that directly minimizes the loss w.r.t. the\nbinarized weights. The underlying proximal step has an efficient closed-form\nsolution, and the second-order information can be efficiently obtained from the\nsecond moments already computed by the Adam optimizer. Experiments on both\nfeedforward and recurrent networks show that the proposed loss-aware\nbinarization algorithm outperforms existing binarization schemes, and is also\nmore robust for wide and deep networks.","url_abs":"http://arxiv.org/abs/1611.01600v3","url_pdf":"http://arxiv.org/pdf/1611.01600v3.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":"loss-aware-binarization-of-deep-networks","repo_url":"https://github.com/houlu369/Loss-aware-Binarization","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":{"status":"ok","spdx":"GPL-3.0"}}],"tasks":[{"task_slug":"binarization","task_name":"Binarization"}],"methods":[{"method_slug":"adam","method_name":"Adam"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=1611.01600","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}