English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

We present a new dataset condensation framework termed Squeeze, Recover andRelabel (SRe$^2$L) that decouples the bilevel optimization of model andsynthetic data during training, to handle varying scales of datasets, modelarchitectures and image resolutions for efficient dataset condensation. Theproposed method demonstrates flexibility across diverse dataset scales andexhibits multiple advantages in terms of arbitrary resolutions of synthesizedimages, low training cost and memory consumption with high-resolutionsynthesis, and the ability to scale up to arbitrary evaluation networkarchitectures. Extensive experiments are conducted on Tiny-ImageNet and fullImageNet-1K datasets. Under 50 IPC, our approach achieves the highest 42.5% and60.8% validation accuracy on Tiny-ImageNet and ImageNet-1K, outperforming allprevious state-of-the-art methods by margins of 14.5% and 32.9%, respectively.Our approach also surpasses MTT in terms of speed by approximately 52$\times$(ConvNet-4) and 16$\times$ (ResNet-18) faster with less memory consumption of11.6$\times$ and 6.4$\times$ during data synthesis. Our code and condenseddatasets of 50, 200 IPC with 4K recovery budget are available athttps://github.com/VILA-Lab/SRe2L.

Squeeze, Recover and Relabel: Dataset Condensation at ImageNet Scale From A New Perspective