Adding a supervised SimCLR-style contrastive loss to DATM trajectory matching improves distilled-dataset accuracy at one image per class on CIFAR-10, CIFAR-100, and Tiny-ImageNet.
In: Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (2022)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Contrastive Learning-Enhanced Trajectory Matching for Small-Scale Dataset Distillation
Adding a supervised SimCLR-style contrastive loss to DATM trajectory matching improves distilled-dataset accuracy at one image per class on CIFAR-10, CIFAR-100, and Tiny-ImageNet.