{"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/harmonic-networks-with-limited-training","title":"Harmonic Networks with Limited Training Samples","arxiv_id":"1905.00135","date":"2019-04-30","proceeding":null,"authors":["Matej Ulicny","Vladimir A. Krylov","Rozenn Dahyot"],"abstract":"Convolutional neural networks (CNNs) are very popular nowadays for image\nprocessing. CNNs allow one to learn optimal filters in a (mostly) supervised\nmachine learning context. However this typically requires abundant labelled\ntraining data to estimate the filter parameters. Alternative strategies have\nbeen deployed for reducing the number of parameters and / or filters to be\nlearned and thus decrease overfitting. In the context of reverting to preset\nfilters, we propose here a computationally efficient harmonic block that uses\nDiscrete Cosine Transform (DCT) filters in CNNs. In this work we examine the\nperformance of harmonic networks in limited training data scenario. We validate\nexperimentally that its performance compares well against scattering networks\nthat use wavelets as preset filters.","url_abs":"http://arxiv.org/abs/1905.00135v1","url_pdf":"http://arxiv.org/pdf/1905.00135v1.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":"harmonic-networks-with-limited-training","repo_url":"https://github.com/matej-ulicny/harmonic-networks","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"BSD-3-Clause"}}],"tasks":[{"task_slug":"image-classification","task_name":"Image Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/image-classification-on-stl-10","task":"Image Classification","dataset":"STL-10","model":"Harmonic WRN-16-8","rank_in_archive_order":33,"of":117,"metrics":{"Percentage correct":"90.45"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}