{"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/training-faster-by-separating-modes-of","title":"Training Faster by Separating Modes of Variation in Batch-normalized Models","arxiv_id":"1806.02892","date":"2018-06-07","proceeding":null,"authors":["Mahdi M. Kalayeh","Mubarak Shah"],"abstract":"Batch Normalization (BN) is essential to effectively train state-of-the-art\ndeep Convolutional Neural Networks (CNN). It normalizes inputs to the layers\nduring training using the statistics of each mini-batch. In this work, we study\nBN from the viewpoint of Fisher kernels. We show that assuming samples within a\nmini-batch are from the same probability density function, then BN is identical\nto the Fisher vector of a Gaussian distribution. That means BN can be explained\nin terms of kernels that naturally emerge from the probability density function\nof the underlying data distribution. However, given the rectifying\nnon-linearities employed in CNN architectures, distribution of inputs to the\nlayers show heavy tail and asymmetric characteristics. Therefore, we propose\napproximating underlying data distribution not with one, but a mixture of\nGaussian densities. Deriving Fisher vector for a Gaussian Mixture Model (GMM),\nreveals that BN can be improved by independently normalizing with respect to\nthe statistics of disentangled sub-populations. We refer to our proposed soft\npiecewise version of BN as Mixture Normalization (MN). Through extensive set of\nexperiments on CIFAR-10 and CIFAR-100, we show that MN not only effectively\naccelerates training image classification and Generative Adversarial networks,\nbut also reaches higher quality models.","url_abs":"http://arxiv.org/abs/1806.02892v2","url_pdf":"http://arxiv.org/pdf/1806.02892v2.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":"training-faster-by-separating-modes-of","repo_url":"https://github.com/b-faye/unsupervised-context-normalization","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}}],"tasks":[{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"image-classification","task_name":"image-classification"}],"methods":[{"method_slug":"mixture-normalization","method_name":"Mixture Normalization"}],"datasets_introduced":[],"methods_introduced":[{"slug":"mixture-normalization","name":"Mixture Normalization","full_name":"Mixture Normalization"}],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1806.02892","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}