{"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/generalized-parametric-contrastive-learning","title":"Generalized Parametric Contrastive Learning","arxiv_id":"2209.12400","date":"2022-09-26","proceeding":null,"authors":["Jiequan Cui","Zhisheng Zhong","Zhuotao Tian","Shu Liu","Bei Yu","Jiaya Jia"],"abstract":"In this paper, we propose the Generalized Parametric Contrastive Learning (GPaCo/PaCo) which works well on both imbalanced and balanced data. Based on theoretical analysis, we observe that supervised contrastive loss tends to bias high-frequency classes and thus increases the difficulty of imbalanced learning. We introduce a set of parametric class-wise learnable centers to rebalance from an optimization perspective. Further, we analyze our GPaCo/PaCo loss under a balanced setting. Our analysis demonstrates that GPaCo/PaCo can adaptively enhance the intensity of pushing samples of the same class close as more samples are pulled together with their corresponding centers and benefit hard example learning. Experiments on long-tailed benchmarks manifest the new state-of-the-art for long-tailed recognition. On full ImageNet, models from CNNs to vision transformers trained with GPaCo loss show better generalization performance and stronger robustness compared with MAE models. Moreover, GPaCo can be applied to the semantic segmentation task and obvious improvements are observed on the 4 most popular benchmarks. Our code is available at https://github.com/dvlab-research/Parametric-Contrastive-Learning.","url_abs":"https://arxiv.org/abs/2209.12400v2","url_pdf":"https://arxiv.org/pdf/2209.12400v2.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":"generalized-parametric-contrastive-learning","repo_url":"https://github.com/dvlab-research/parametric-contrastive-learning","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"generalized-parametric-contrastive-learning","repo_url":"https://github.com/dvlab-research/imbalanced-learning","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"generalized-parametric-contrastive-learning","repo_url":"https://github.com/dvlab-research/rescom","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"generalized-parametric-contrastive-learning","repo_url":"https://github.com/jiequancui/Parametric-Contrastive-Learning","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"contrastive-learning","task_name":"Contrastive Learning"},{"task_slug":"domain-generalization","task_name":"Domain Generalization"},{"task_slug":"image-classification","task_name":"Image Classification"},{"task_slug":"long-tail-learning","task_name":"Long-tail Learning"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[{"method_slug":"contrastive-learning","method_name":"Contrastive Learning"},{"method_slug":"mae","method_name":"MAE"},{"method_slug":"supervised-contrastive-loss","method_name":"Supervised Contrastive Loss"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/domain-generalization-on-imagenet-c","task":"Domain Generalization","dataset":"ImageNet-C","model":"GPaCo (ViT-L)","rank_in_archive_order":13,"of":47,"metrics":{"mean Corruption Error (mCE)":"39.0"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-imagenet-r","task":"Domain Generalization","dataset":"ImageNet-R","model":"GPaCo (ViT-L)","rank_in_archive_order":15,"of":39,"metrics":{"Top-1 Error Rate":"39.7"},"uses_additional_data":false},{"leaderboard":"/sota/domain-generalization-on-imagenet-sketch","task":"Domain Generalization","dataset":"ImageNet-Sketch","model":"GPaCo (ViT-L)","rank_in_archive_order":12,"of":20,"metrics":{"Top-1 accuracy":"48.3"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-imagenet","task":"Image Classification","dataset":"ImageNet","model":"GPaCo (ViT-L)","rank_in_archive_order":176,"of":1060,"metrics":{"Top 1 Accuracy":"86.01%"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-imagenet","task":"Image Classification","dataset":"ImageNet","model":"GPaCo (Vit-B)","rank_in_archive_order":361,"of":1060,"metrics":{"Top 1 Accuracy":"84.0%"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-imagenet","task":"Image Classification","dataset":"ImageNet","model":"GPaCo (ResNet-50)","rank_in_archive_order":745,"of":1060,"metrics":{"Top 1 Accuracy":"79.7%"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-inaturalist-2018","task":"Image Classification","dataset":"iNaturalist 2018","model":"GPaCo (ResNet-152)","rank_in_archive_order":21,"of":60,"metrics":{"Top-1 Accuracy":"78.1%"},"uses_additional_data":false},{"leaderboard":"/sota/image-classification-on-inaturalist-2018","task":"Image Classification","dataset":"iNaturalist 2018","model":"GPaCo (ResNet-50)","rank_in_archive_order":25,"of":60,"metrics":{"Top-1 Accuracy":"75.4%"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-imagenet-lt","task":"Long-tail Learning","dataset":"ImageNet-LT","model":"GPaCo (2-ResNeXt101-32x4d)","rank_in_archive_order":11,"of":69,"metrics":{"Top-1 Accuracy":"63.2"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-places-lt","task":"Long-tail Learning","dataset":"Places-LT","model":"GPaCo (ResNet-152)","rank_in_archive_order":13,"of":29,"metrics":{"Top-1 Accuracy":"41.7"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-inaturalist-2018","task":"Long-tail Learning","dataset":"iNaturalist 2018","model":"GPaCo (2-R152)","rank_in_archive_order":7,"of":43,"metrics":{"Top-1 Accuracy":"79.8%"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-inaturalist-2018","task":"Long-tail Learning","dataset":"iNaturalist 2018","model":"GPaCo (ResNet-152)","rank_in_archive_order":8,"of":43,"metrics":{"Top-1 Accuracy":"78.1%"},"uses_additional_data":false},{"leaderboard":"/sota/long-tail-learning-on-inaturalist-2018","task":"Long-tail Learning","dataset":"iNaturalist 2018","model":"GPaCo (ResNet-50)","rank_in_archive_order":12,"of":43,"metrics":{"Top-1 Accuracy":"75.4%"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-ade20k","task":"Semantic Segmentation","dataset":"ADE20K","model":"GPaCo (Swin-L)","rank_in_archive_order":64,"of":235,"metrics":{"Validation mIoU":"54.3"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-pascal-context","task":"Semantic Segmentation","dataset":"PASCAL Context","model":"GPaCo (ResNet101)","rank_in_archive_order":25,"of":66,"metrics":{"mIoU":"56.2"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2209.12400","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}