{"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/competitive-multi-scale-convolution","title":"Competitive Multi-scale Convolution","arxiv_id":"1511.05635","date":"2015-11-18","proceeding":null,"authors":["Zhibin Liao","Gustavo Carneiro"],"abstract":"In this paper, we introduce a new deep convolutional neural network (ConvNet)\nmodule that promotes competition among a set of multi-scale convolutional\nfilters. This new module is inspired by the inception module, where we replace\nthe original collaborative pooling stage (consisting of a concatenation of the\nmulti-scale filter outputs) by a competitive pooling represented by a maxout\nactivation unit. This extension has the following two objectives: 1) the\nselection of the maximum response among the multi-scale filters prevents filter\nco-adaptation and allows the formation of multiple sub-networks within the same\nmodel, which has been shown to facilitate the training of complex learning\nproblems; and 2) the maxout unit reduces the dimensionality of the outputs from\nthe multi-scale filters. We show that the use of our proposed module in typical\ndeep ConvNets produces classification results that are either better than or\ncomparable to the state of the art on the following benchmark datasets: MNIST,\nCIFAR-10, CIFAR-100 and SVHN.","url_abs":"http://arxiv.org/abs/1511.05635v1","url_pdf":"http://arxiv.org/pdf/1511.05635v1.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":[],"tasks":[{"task_slug":"image-classification","task_name":"Image Classification"}],"methods":[{"method_slug":"maxout","method_name":"Maxout"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-classification-on-cifar-10","task":"Image Classification","dataset":"CIFAR-10","model":"CMsC","rank_in_archive_order":177,"of":265,"metrics":{"Percentage correct":"93.1"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-cifar-100","task":"Image Classification","dataset":"CIFAR-100","model":"CMsC","rank_in_archive_order":168,"of":211,"metrics":{"Percentage correct":"72.4"},"uses_additional_data":true},{"leaderboard":"/sota/image-classification-on-mnist","task":"Image Classification","dataset":"MNIST","model":"CMsC","rank_in_archive_order":20,"of":81,"metrics":{"Percentage error":"0.3"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-svhn","task":"Image Classification","dataset":"SVHN","model":"CMsC","rank_in_archive_order":23,"of":62,"metrics":{"Percentage error":"1.8"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1511.05635","atlas_url":"https://app.syntology.ai/?focus=1511.05635","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}