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

Neural networks are prone to learn easy solutions from superficial statisticsin the data, namely shortcut learning, which impairs generalization androbustness of models. We propose a data augmentation strategy, named DFM-X,that leverages knowledge about frequency shortcuts, encoded in DominantFrequencies Maps computed for image classification models. We randomly selectX% training images of certain classes for augmentation, and process them byretaining the frequencies included in the DFMs of other classes. This strategycompels the models to leverage a broader range of frequencies forclassification, rather than relying on specific frequency sets. Thus, themodels learn more deep and task-related semantics compared to their counterparttrained with standard setups. Unlike other commonly used augmentationtechniques which focus on increasing the visual variations of training data,our method targets exploiting the original data efficiently, by distillingprior knowledge about destructive learning behavior of models from data. Ourexperimental results demonstrate that DFM-X improves robustness against commoncorruptions and adversarial attacks. It can be seamlessly integrated with otheraugmentation techniques to further enhance the robustness of models.

DFM-X: Augmentation by Leveraging Prior Knowledge of Shortcut Learning