{"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/a-new-loss-function-for-cnn-classifier-based","title":"A New Loss Function for CNN Classifier Based on Pre-defined Evenly-Distributed Class Centroids","arxiv_id":"1904.06008","date":"2019-04-12","proceeding":null,"authors":["Qiuyu Zhu","Pengju Zhang","Xin Ye"],"abstract":"With the development of convolutional neural networks (CNNs) in recent years,\nthe network structure has become more and more complex and varied, and has\nachieved very good results in pattern recognition, image classification, object\ndetection and tracking. For CNNs used for image classification, in addition to\nthe network structure, more and more research is now focusing on the\nimprovement of the loss function, so as to enlarge the inter-class feature\ndifferences, and reduce the intra-class feature variations as soon as possible.\nBesides the traditional Softmax, typical loss functions include L-Softmax,\nAM-Softmax, ArcFace, and Center loss, etc. Based on the concept of predefined\nevenly-distributed class centroids (PEDCC) in CSAE network, this paper proposes\na PEDCC-based loss function called PEDCC-Loss, which can make the inter-class\ndistance maximal and intra-class distance small enough in hidden feature space.\nMultiple experiments on image classification and face recognition have proved\nthat our method achieve the best recognition accuracy, and network training is\nstable and easy to converge. Code is available in\nhttps://github.com/ZLeopard/PEDCC-Loss","url_abs":"http://arxiv.org/abs/1904.06008v2","url_pdf":"http://arxiv.org/pdf/1904.06008v2.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":"a-new-loss-function-for-cnn-classifier-based","repo_url":"https://github.com/ZLeopard/PEDCC-Loss","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"face-recognition","task_name":"Face Recognition"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"object-detection","task_name":"Object Detection"},{"task_slug":"image-classification","task_name":"image-classification"},{"task_slug":"object-detection-1","task_name":"object-detection"}],"methods":[{"method_slug":"arcface","method_name":"ArcFace"},{"method_slug":"softmax","method_name":"Softmax"}],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1904.06008","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}