{"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/solving-internal-covariate-shift-in-deep","title":"Solving internal covariate shift in deep learning with linked neurons","arxiv_id":"1712.02609","date":"2017-12-07","proceeding":null,"authors":["Carles Roger Riera Molina","Oriol Pujol Vila"],"abstract":"This work proposes a novel solution to the problem of internal covariate\nshift and dying neurons using the concept of linked neurons. We define the\nneuron linkage in terms of two constraints: first, all neuron activations in\nthe linkage must have the same operating point. That is to say, all of them\nshare input weights. Secondly, a set of neurons is linked if and only if there\nis at least one member of the linkage that has a non-zero gradient in regard to\nthe input of the activation function. This means that for any input in the\nactivation function, there is at least one member of the linkage that operates\nin a non-flat and non-zero area. This simple change has profound implications\nin the network learning dynamics. In this article we explore the consequences\nof this proposal and show that by using this kind of units, internal covariate\nshift is implicitly solved. As a result of this, the use of linked neurons\nallows to train arbitrarily large networks without any architectural or\nalgorithmic trick, effectively removing the need of using re-normalization\nschemes such as Batch Normalization, which leads to halving the required\ntraining time. It also solves the problem of the need for standarized input\ndata. Results show that the units using the linkage not only do effectively\nsolve the aforementioned problems, but are also a competitive alternative with\nrespect to state-of-the-art with very promising results.","url_abs":"http://arxiv.org/abs/1712.02609v1","url_pdf":"http://arxiv.org/pdf/1712.02609v1.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":"solving-internal-covariate-shift-in-deep","repo_url":"https://github.com/blauigris/linked_neurons","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[{"task_slug":"deep-learning","task_name":"Deep Learning"}],"methods":[{"method_slug":"batch-normalization","method_name":"Batch Normalization"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}