{"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-quantization-friendly-separable-convolution","title":"A Quantization-Friendly Separable Convolution for MobileNets","arxiv_id":"1803.08607","date":"2018-03-22","proceeding":null,"authors":["Tao Sheng","Chen Feng","Shaojie Zhuo","Xiaopeng Zhang","Liang Shen","Mickey Aleksic"],"abstract":"As deep learning (DL) is being rapidly pushed to edge computing, researchers\ninvented various ways to make inference computation more efficient on\nmobile/IoT devices, such as network pruning, parameter compression, and etc.\nQuantization, as one of the key approaches, can effectively offload GPU, and\nmake it possible to deploy DL on fixed-point pipeline. Unfortunately, not all\nexisting networks design are friendly to quantization. For example, the popular\nlightweight MobileNetV1, while it successfully reduces parameter size and\ncomputation latency with separable convolution, our experiment shows its\nquantized models have large accuracy gap against its float point models. To\nresolve this, we analyzed the root cause of quantization loss and proposed a\nquantization-friendly separable convolution architecture. By evaluating the\nimage classification task on ImageNet2012 dataset, our modified MobileNetV1\nmodel can archive 8-bit inference top-1 accuracy in 68.03%, almost closed the\ngap to the float pipeline.","url_abs":"http://arxiv.org/abs/1803.08607v3","url_pdf":"http://arxiv.org/pdf/1803.08607v3.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-quantization-friendly-separable-convolution","repo_url":"https://github.com/pinto0309/onnx2tf","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":null}],"tasks":[{"task_slug":"edge-computing","task_name":"Edge-computing"},{"task_slug":null,"task_name":"GPU"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"network-pruning","task_name":"Network Pruning"},{"task_slug":"quantization","task_name":"Quantization"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[{"method_slug":"1x1-convolution","method_name":"1x1 Convolution"},{"method_slug":"average-pooling","method_name":"Average Pooling"},{"method_slug":"batch-normalization","method_name":"Batch Normalization"},{"method_slug":"convolution","method_name":"Convolution"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"depthwise-convolution","method_name":"Depthwise Convolution"},{"method_slug":"depthwise-separable-convolution","method_name":"Depthwise Separable Convolution"},{"method_slug":"global-average-pooling","method_name":"Global Average Pooling"},{"method_slug":"mobilenetv1","method_name":"MobileNetV1"},{"method_slug":"pointwise-convolution","method_name":"Pointwise Convolution"},{"method_slug":"relu","method_name":"ReLU"},{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1803.08607","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}