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

Machine learning models are used in many sensitive areas where besidespredictive accuracy their comprehensibility is also important. Interpretabilityof prediction models is necessary to determine their biases and causes oferrors, and is a necessary prerequisite for users' confidence. For complexstate-of-the-art black-box models post-hoc model-independent explanationtechniques are an established solution. Popular and effective techniques, suchas IME, LIME, and SHAP, use perturbation of instance features to explainindividual predictions. Recently, Slack et al. (2020) put their robustness intoquestion by showing that their outcomes can be manipulated due to poorperturbation sampling employed. This weakness would allow dieselgate typecheating of owners of sensitive models who could deceive inspection and hidepotentially unethical or illegal biases existing in their predictive models.This could undermine public trust in machine learning models and give rise tolegal restrictions on their use.  We show that better sampling in these explanation methods prevents maliciousmanipulations. The proposed sampling uses data generators that learn thetraining set distribution and generate new perturbation instances much moresimilar to the training set. We show that the improved sampling increases therobustness of the LIME and SHAP, while previously untested method IME isalready the most robust of all.

Better sampling in explanation methods can prevent dieselgate-like deception