CCL: Class-wise curriculum learning for class imbalance problems
Entity
UAM. Departamento de Tecnología Electrónica y de las ComunicacionesPublisher
IEEEDate
2022-10-18Citation
10.1109/ICIP46576.2022.9897273
IIEEE International Conference on Image Processing (ICIP), Bordeaux, France, 2022
ISSN
2381-8549ISBN
978-1-6654-9620-9DOI
10.1109/ICIP46576.2022.9897273Funded by
Partially founded by the Comunidad de Madrid through the SI1/PJI/2019-00414 AISEEME projectProject
Comunidad de Madrid. SI1/PJI/2019-00414Editor's Version
https://doi.org/10.1109/ICIP46576.2022.9897273Subjects
Class imbalance; Curriculum learning; Image Classification; Sample scoring; Training pace; TelecomunicacionesRights
© 2022 IEEEAbstract
Computer vision datasets usually present long-tailed training distributions where the classes are not represented with the same number of training samples. This so-called class imbalance problem hinders the proper learning of inference models, biasing them towards over-represented classes and decreasing their generalization. Adopted solutions to tackle the effect of class imbalance are based on weighting the training loss according to the number of class samples, leading to regimes where low-represented classes guide the learning just accounting for their cardinal number. To also incorporate class complexity in the process, we propose a novel training scheme called CCL: Class-wise Curriculum Learning. Classes are first sorted based on a difficulty criterion which not only accounts for the number of training samples but also for their training outcomes. The curriculum is then used to guide the training: easy classes are fed first and-incrementally, the more difficult ones are added. The proposed approach is validated for image classification using long-tailed datasets. Results show that when the proposed Class-wise Curriculum Learning scheme is used, trained models outperform specific state-of-the-art methods devoted to handle the class imbalance problem. The code, data and reported models described along this paper are publicly available at https://github.com/vpulab/CCL
Files in this item
Google Scholar:Escudero Viñolo, Marcos
-
López Cifuentes, Alejandro
This item appears in the following Collection(s)
Related items
Showing items related by title, author, creator and subject.