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/

Zendy Plus

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

Zendy Tools

Zendy Open

Data-Efficient GANs (DE-GANs), which aim to learn generative models with alimited amount of training data, encounter several challenges for generatinghigh-quality samples. Since data augmentation strategies have largelyalleviated the training instability, how to further improve the generativeperformance of DE-GANs becomes a hotspot. Recently, contrastive learning hasshown the great potential of increasing the synthesis quality of DE-GANs, yetrelated principles are not well explored. In this paper, we revisit and comparedifferent contrastive learning strategies in DE-GANs, and identify (i) thecurrent bottleneck of generative performance is the discontinuity of latentspace; (ii) compared to other contrastive learning strategies,Instance-perturbation works towards latent space continuity, which brings themajor improvement to DE-GANs. Based on these observations, we propose FakeCLR,which only applies contrastive learning on perturbed fake samples, and devisesthree related training techniques: Noise-related Latent Augmentation,Diversity-aware Queue, and Forgetting Factor of Queue. Our experimental resultsmanifest the new state of the arts on both few-shot generation and limited-datageneration. On multiple datasets, FakeCLR acquires more than 15% FIDimprovement compared to existing DE-GANs. Code is available athttps://github.com/iceli1007/FakeCLR.

FakeCLR: Exploring Contrastive Learning for Solving Latent Discontinuity in Data-Efficient GANs