{"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/convolutional-kernel-networks","title":"Convolutional Kernel Networks","arxiv_id":"1406.3332","date":"2014-06-12","proceeding":"NeurIPS 2014 12","authors":["Julien Mairal","Piotr Koniusz","Zaid Harchaoui","Cordelia Schmid"],"abstract":"An important goal in visual recognition is to devise image representations\nthat are invariant to particular transformations. In this paper, we address\nthis goal with a new type of convolutional neural network (CNN) whose\ninvariance is encoded by a reproducing kernel. Unlike traditional approaches\nwhere neural networks are learned either to represent data or for solving a\nclassification task, our network learns to approximate the kernel feature map\non training data. Such an approach enjoys several benefits over classical ones.\nFirst, by teaching CNNs to be invariant, we obtain simple network architectures\nthat achieve a similar accuracy to more complex ones, while being easy to train\nand robust to overfitting. Second, we bridge a gap between the neural network\nliterature and kernels, which are natural tools to model invariance. We\nevaluate our methodology on visual recognition tasks where CNNs have proven to\nperform well, e.g., digit recognition with the MNIST dataset, and the more\nchallenging CIFAR-10 and STL-10 datasets, where our accuracy is competitive\nwith the state of the art.","url_abs":"http://arxiv.org/abs/1406.3332v2","url_pdf":"http://arxiv.org/pdf/1406.3332v2.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":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-classification-on-cifar-10","task":"Image Classification","dataset":"CIFAR-10","model":"CKN","rank_in_archive_order":243,"of":265,"metrics":{"Percentage correct":"82.2"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-mnist","task":"Image Classification","dataset":"MNIST","model":"CKN","rank_in_archive_order":25,"of":81,"metrics":{"Percentage error":"0.4"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-stl-10","task":"Image Classification","dataset":"STL-10","model":"CKN","rank_in_archive_order":101,"of":117,"metrics":{"Percentage correct":"62.3"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1406.3332","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}