'Diachronic interpretability and machine learning systems' by Sylvie Delacroix

Diachronic interpretability and machine learning systems

In our seventh online-first article (January 2022), Sylvie Delacroix analyses how the passage of time affects the interpretability of machine learning systems, arguing that 'ensemble contestability features' are needed to enable contestation across temporal contexts.

Reply by Zachary C. Lipton, Carnegie Mellon University.