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

Transformer-based models have gained popularity in the field of naturallanguage processing (NLP) and are extensively utilized in computer vision tasksand multi-modal models such as GPT4. This paper presents a novel method toenhance the explainability of Transformer-based image classification models.Our method aims to improve trust in classification results and empower users togain a deeper understanding of the model for downstream tasks by providingvisualizations of class-specific maps. We introduce two modules: the``Relationship Weighted Out" and the ``Cut" modules. The ``RelationshipWeighted Out" module focuses on extracting class-specific information fromintermediate layers, enabling us to highlight relevant features. Additionally,the ``Cut" module performs fine-grained feature decomposition, taking intoaccount factors such as position, texture, and color. By integrating thesemodules, we generate dense class-specific visual explainability maps. Wevalidate our method with extensive qualitative and quantitative experiments onthe ImageNet dataset. Furthermore, we conduct a large number of experiments onthe LRN dataset, specifically designed for automatic driving danger alerts, toevaluate the explainability of our method in complex backgrounds. The resultsdemonstrate a significant improvement over previous methods. Moreover, weconduct ablation experiments to validate the effectiveness of each module.Through these experiments, we are able to confirm the respective contributionsof each module, thus solidifying the overall effectiveness of our proposedapproach.

R-Cut: Enhancing Explainability in Vision Transformers with Relationship Weighted Out and Cut