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

Traditional tabular classifiers provide explainable decision-making withinterpretable features(concepts). However, using their explainability in visiontasks has been limited due to the pixel representation of images. In thispaper, we design Img2Tabs that classify images by concepts to harness theexplainability of tabular classifiers. Img2Tabs encode image pixels intotabular features by StyleGAN inversion. Since not all of the resulting featuresare class-relevant or interpretable due to their generative nature, we expectImg2Tab classifiers to discover class-relevant concepts automatically from theStyleGAN features. Thus, we propose a novel method using the Wasserstein-1metric to quantify class-relevancy and interpretability simultaneously. Usingthis method, we investigate whether important features extracted by tabularclassifiers are class-relevant concepts. Consequently, we determine the mosteffective classifier for Img2Tabs in terms of discovering class-relevantconcepts automatically from StyleGAN features. In evaluations, we demonstrateconcept-based explanations through importance and visualization. Img2Tabachieves top-1 accuracy that is on par with CNN classifiers and deep featurelearning baselines. Additionally, we show that users can easily debug Img2Tabclassifiers at the concept level to ensure unbiased and fair decision-makingwithout sacrificing accuracy.

Img2Tab: Automatic Class Relevant Concept Discovery from StyleGAN Features for Explainable Image Classification