Contrastive Learning for Online Semi-Supervised General Continual Learning

Nicolas Michel (1) , Romain Negrel (1) , Giovanni Chierchia (1) , Jean-François Bercher (1)

1 LIGM - Laboratoire d'Informatique Gaspard-Monge

We study Online Continual Learning with missing labels and propose SemiCon, a new contrastive loss designed for partly labeled data. We demonstrate its efficiency by devising a memory-based method trained on an unlabeled data stream, where every data added to memory is labeled using an oracle. Our approach outperforms existing semi-supervised methods when few labels are available, and obtain similar results to state-of-the-art supervised methods while using only 2.6% of labels on Split-CIFAR10 and 10% of labels on Split-CIFAR100.

Mots clés: Continual Learning; Contrastive Learning; Semi-Supervised Learning; Memory; Continual Learning

Domaines: Informatique [cs]

Nicolas Michel, Romain Negrel, Giovanni Chierchia, Jean-François Bercher. Contrastive Learning for Online Semi-Supervised General Continual Learning. 2022 IEEE International Conference on Image Processing (ICIP), Oct 2022, Bordeaux, France. pp.1896-1900, ⟨10.1109/ICIP46576.2022.9897290⟩. ⟨hal-03866436⟩