Reproduce “We Need Explanation Cards to Connect Explanation Algorithms to the Real World” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Eric Günther, Balázs Szabados, Kristof Meding, Gunnar König, Sebastian Bordt, Ulrike von Luxburg
The authors propose Explanation Cards that pair post-hoc explanations with checks of their validity or robustness and guidance on what recipients may infer. For counterfactuals, stability can make an explanation informative under stated assumptions; for SHAP, local stability and additivity conditions govern when attributions track prediction changes. Worked tabular examples illustrate these designs, but no user study establishes reduced misinterpretation or intervention success.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.