{"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/winner-take-all-autoencoders","title":"Winner-Take-All Autoencoders","arxiv_id":"1409.2752","date":"2014-09-09","proceeding":"NeurIPS 2015 12","authors":["Alireza Makhzani","Brendan Frey"],"abstract":"In this paper, we propose a winner-take-all method for learning hierarchical\nsparse representations in an unsupervised fashion. We first introduce\nfully-connected winner-take-all autoencoders which use mini-batch statistics to\ndirectly enforce a lifetime sparsity in the activations of the hidden units. We\nthen propose the convolutional winner-take-all autoencoder which combines the\nbenefits of convolutional architectures and autoencoders for learning\nshift-invariant sparse representations. We describe a way to train\nconvolutional autoencoders layer by layer, where in addition to lifetime\nsparsity, a spatial sparsity within each feature map is achieved using\nwinner-take-all activation functions. We will show that winner-take-all\nautoencoders can be used to to learn deep sparse representations from the\nMNIST, CIFAR-10, ImageNet, Street View House Numbers and Toronto Face datasets,\nand achieve competitive classification performance.","url_abs":"http://arxiv.org/abs/1409.2752v2","url_pdf":"http://arxiv.org/pdf/1409.2752v2.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":"winner-take-all-autoencoders","repo_url":"https://github.com/Kaixhin/Autoencoders","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"torch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"all","task_name":"All"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1409.2752","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}