{"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/towards-optimal-structured-cnn-pruning-via","title":"Towards Optimal Structured CNN Pruning via Generative Adversarial Learning","arxiv_id":"1903.09291","date":"2019-03-22","proceeding":"CVPR 2019 6","authors":["Shaohui Lin","Rongrong Ji","Chenqian Yan","Baochang Zhang","Liujuan Cao","Qixiang Ye","Feiyue Huang","David Doermann"],"abstract":"Structured pruning of filters or neurons has received increased focus for\ncompressing convolutional neural networks. Most existing methods rely on\nmulti-stage optimizations in a layer-wise manner for iteratively pruning and\nretraining which may not be optimal and may be computation intensive. Besides,\nthese methods are designed for pruning a specific structure, such as filter or\nblock structures without jointly pruning heterogeneous structures. In this\npaper, we propose an effective structured pruning approach that jointly prunes\nfilters as well as other structures in an end-to-end manner. To accomplish\nthis, we first introduce a soft mask to scale the output of these structures by\ndefining a new objective function with sparsity regularization to align the\noutput of baseline and network with this mask. We then effectively solve the\noptimization problem by generative adversarial learning (GAL), which learns a\nsparse soft mask in a label-free and an end-to-end manner. By forcing more\nscaling factors in the soft mask to zero, the fast iterative\nshrinkage-thresholding algorithm (FISTA) can be leveraged to fast and reliably\nremove the corresponding structures. Extensive experiments demonstrate the\neffectiveness of GAL on different datasets, including MNIST, CIFAR-10 and\nImageNet ILSVRC 2012. For example, on ImageNet ILSVRC 2012, the pruned\nResNet-50 achieves 10.88\\% Top-5 error and results in a factor of 3.7x speedup.\nThis significantly outperforms state-of-the-art methods.","url_abs":"http://arxiv.org/abs/1903.09291v1","url_pdf":"http://arxiv.org/pdf/1903.09291v1.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":"towards-optimal-structured-cnn-pruning-via","repo_url":"https://github.com/ShaohuiLin/GAL","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[],"methods":[{"method_slug":"pruning","method_name":"Pruning"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1903.09291","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}