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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Methods for model explainability have become increasingly critical fortesting the fairness and soundness of deep learning. Concept-basedinterpretability techniques, which use a small set of human-interpretableconcept exemplars in order to measure the influence of a concept on a model'sinternal representation of input, are an important thread in this line ofresearch. In this work we show that these explainability methods can suffer thesame vulnerability to adversarial attacks as the models they are meant toanalyze. We demonstrate this phenomenon on two well-known concept-basedinterpretability methods: TCAV and faceted feature visualization. We show thatby carefully perturbing the examples of the concept that is being investigated,we can radically change the output of the interpretability method. The attacksthat we propose can either induce positive interpretations (polka dots are animportant concept for a model when classifying zebras) or negativeinterpretations (stripes are not an important factor in identifying images of azebra). Our work highlights the fact that in safety-critical applications,there is need for security around not only the machine learning pipeline butalso the model interpretation process.

Making Corgis Important for Honeycomb Classification: Adversarial Attacks on Concept-based Explainability Tools