{"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/perturbative-neural-networks","title":"Perturbative Neural Networks","arxiv_id":"1806.01817","date":"2018-06-05","proceeding":"CVPR 2018 6","authors":["Felix Juefei-Xu","Vishnu Naresh Boddeti","Marios Savvides"],"abstract":"Convolutional neural networks are witnessing wide adoption in computer vision\nsystems with numerous applications across a range of visual recognition tasks.\nMuch of this progress is fueled through advances in convolutional neural\nnetwork architectures and learning algorithms even as the basic premise of a\nconvolutional layer has remained unchanged. In this paper, we seek to revisit\nthe convolutional layer that has been the workhorse of state-of-the-art visual\nrecognition models. We introduce a very simple, yet effective, module called a\nperturbation layer as an alternative to a convolutional layer. The perturbation\nlayer does away with convolution in the traditional sense and instead computes\nits response as a weighted linear combination of non-linearly activated\nadditive noise perturbed inputs. We demonstrate both analytically and\nempirically that this perturbation layer can be an effective replacement for a\nstandard convolutional layer. Empirically, deep neural networks with\nperturbation layers, called Perturbative Neural Networks (PNNs), in lieu of\nconvolutional layers perform comparably with standard CNNs on a range of visual\ndatasets (MNIST, CIFAR-10, PASCAL VOC, and ImageNet) with fewer parameters.","url_abs":"http://arxiv.org/abs/1806.01817v1","url_pdf":"http://arxiv.org/pdf/1806.01817v1.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":"perturbative-neural-networks","repo_url":"https://github.com/fengjiqiang/pretrainedmodel_pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"perturbative-neural-networks","repo_url":"https://github.com/juefeix/pnn.pytorch","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"gone","observed_at":"2026-09-18","how":"tree_404+repo_404"}},{"paper_slug":"perturbative-neural-networks","repo_url":"https://github.com/juefeix/pnn.pytorch.update","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"gone","observed_at":"2026-09-18","how":"tree_404+repo_404"}}],"tasks":[],"methods":[{"method_slug":"convolution","method_name":"Convolution"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}