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

Domain generalization (DG) intends to train a model on multiple sourcedomains to ensure that it can generalize well to an arbitrary unseen targetdomain. The acquisition of domain-invariant representations is pivotal for DGas they possess the ability to capture the inherent semantic information of thedata, mitigate the influence of domain shift, and enhance the generalizationcapability of the model. Adopting multiple perspectives, such as the sample andthe feature, proves to be effective. The sample perspective facilitates dataaugmentation through data manipulation techniques, whereas the featureperspective enables the extraction of meaningful generalization features. Inthis paper, we focus on improving the generalization ability of the model bycompelling it to acquire domain-invariant representations from both the sampleand feature perspectives by disentangling spurious correlations and enhancingpotential correlations. 1) From the sample perspective, we develop a frequencyrestriction module, guiding the model to focus on the relevant correlationsbetween object features and labels, thereby disentangling spuriouscorrelations. 2) From the feature perspective, the simple Tail Interactionmodule implicitly enhances potential correlations among all samples from allsource domains, facilitating the acquisition of domain-invariantrepresentations across multiple domains for the model. The experimental resultsshow that Convolutional Neural Networks (CNNs) or Multi-Layer Perceptrons(MLPs) with a strong baseline embedded with these two modules can achievesuperior results, e.g., an average accuracy of 92.30% on Digits-DG.

Learning Generalizable Models via Disentangling Spurious and Enhancing Potential Correlations