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

Learning with noisy labels has been studied to address incorrect labelannotations in real-world applications. In this paper, we present ChiMera, atwo-stage learning-from-noisy-labels framework based on semi-supervisedlearning, developed based on a novel contrastive learning technique MixCLR. Thekey idea of MixCLR is to learn and refine the representations of mixedaugmentations from two different images to better resist label noise. ChiMerajointly learns the representations of the original data distribution andmixed-up data distribution via MixCLR, introducing many additional augmentedsamples to fill in the gap between different classes. This results in a moresmoothed representation space learned by contrastive learning with betteralignment and a more robust decision boundary. By exploiting MixCLR, ChiMeraalso improves the label diffusion process in the semi-supervised noise recoverystage and further boosts its ability to diffuse correct label information. Weevaluated ChiMera on seven real-world datasets and obtained state-of-the-artperformance on both symmetric noise and asymmetric noise. Our method opens upnew avenues for using contrastive learning on learning with noisy labels and weenvision MixCLR to be broadly applicable to other applications.

ChiMera: Learning with noisy labels by contrasting mixed-up augmentations