{"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/a-framework-for-fast-and-efficient-neural","title":"Efficient Neural Network Compression","arxiv_id":"1811.12781","date":"2018-11-30","proceeding":"CVPR 2019 6","authors":["Hyeji Kim","Muhammad Umar Karim Khan","Chong-Min Kyung"],"abstract":"Network compression reduces the computational complexity and memory\nconsumption of deep neural networks by reducing the number of parameters. In\nSVD-based network compression, the right rank needs to be decided for every\nlayer of the network. In this paper, we propose an efficient method for\nobtaining the rank configuration of the whole network. Unlike previous methods\nwhich consider each layer separately, our method considers the whole network to\nchoose the right rank configuration. We propose novel accuracy metrics to\nrepresent the accuracy and complexity relationship for a given neural network.\nWe use these metrics in a non-iterative fashion to obtain the right rank\nconfiguration which satisfies the constraints on FLOPs and memory while\nmaintaining sufficient accuracy. Experiments show that our method provides\nbetter compromise between accuracy and computational complexity/memory\nconsumption while performing compression at much higher speed. For VGG-16 our\nnetwork can reduce the FLOPs by 25% and improve accuracy by 0.7% compared to\nthe baseline, while requiring only 3 minutes on a CPU to search for the right\nrank configuration. Previously, similar results were achieved in 4 hours with 8\nGPUs. The proposed method can be used for lossless compression of a neural\nnetwork as well. The better accuracy and complexity compromise, as well as the\nextremely fast speed of our method makes it suitable for neural network\ncompression.","url_abs":"http://arxiv.org/abs/1811.12781v3","url_pdf":"http://arxiv.org/pdf/1811.12781v3.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":"a-framework-for-fast-and-efficient-neural","repo_url":"https://github.com/Hyeji-Kim/ENC","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"caffe2","reach":null}],"tasks":[{"task_slug":null,"task_name":"CPU"},{"task_slug":"efficient-neural-network","task_name":"Efficient Neural Network"},{"task_slug":"neural-network-compression","task_name":"Neural Network Compression"}],"methods":[{"method_slug":"speed","method_name":"SPEED"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}