{"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/lcnn-lookup-based-convolutional-neural","title":"LCNN: Lookup-based Convolutional Neural Network","arxiv_id":"1611.06473","date":"2016-11-20","proceeding":"CVPR 2017 7","authors":["Hessam Bagherinezhad","Mohammad Rastegari","Ali Farhadi"],"abstract":"Porting state of the art deep learning algorithms to resource constrained\ncompute platforms (e.g. VR, AR, wearables) is extremely challenging. We propose\na fast, compact, and accurate model for convolutional neural networks that\nenables efficient learning and inference. We introduce LCNN, a lookup-based\nconvolutional neural network that encodes convolutions by few lookups to a\ndictionary that is trained to cover the space of weights in CNNs. Training LCNN\ninvolves jointly learning a dictionary and a small set of linear combinations.\nThe size of the dictionary naturally traces a spectrum of trade-offs between\nefficiency and accuracy. Our experimental results on ImageNet challenge show\nthat LCNN can offer 3.2x speedup while achieving 55.1% top-1 accuracy using\nAlexNet architecture. Our fastest LCNN offers 37.6x speed up over AlexNet while\nmaintaining 44.3% top-1 accuracy. LCNN not only offers dramatic speed ups at\ninference, but it also enables efficient training. In this paper, we show the\nbenefits of LCNN in few-shot learning and few-iteration learning, two crucial\naspects of on-device training of deep learning models.","url_abs":"http://arxiv.org/abs/1611.06473v2","url_pdf":"http://arxiv.org/pdf/1611.06473v2.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":"lcnn-lookup-based-convolutional-neural","repo_url":"https://github.com/MindCode-4/code-7/tree/main/lcn-master-ms","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null},{"paper_slug":"lcnn-lookup-based-convolutional-neural","repo_url":"https://github.com/MindCode-4/code-7/tree/main/learning-an-explicit-hyperparameter","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"mindspore","reach":null}],"tasks":[{"task_slug":"few-shot-learning","task_name":"Few-Shot Learning"}],"methods":[{"method_slug":"1x1-convolution","method_name":"1x1 Convolution"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"grouped-convolution","method_name":"Grouped Convolution"},{"method_slug":"local-response-normalization","method_name":"Local Response Normalization"},{"method_slug":"max-pooling","method_name":"Max Pooling"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"speed","method_name":"SPEED"},{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1611.06473","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}