Lei You

DECAF reveals that the trajectory of model responses can provide deeper insights than mere magnitude, with a staggering 96.4% alignment with observed behaviors compared to just 35% for traditional methods.

BoE transforms candidate selection by leveraging partial verification, significantly enhancing outcomes in vision-language tasks where complete evaluations are unattainable.

Distributional counterfactual explanations are now possible for black-box tabular models, thanks to a novel sparse search algorithm that sidesteps the need for gradients.

Escape the curse of off-manifold Shapley values: this new method leverages optimal generative flows to produce attributions that actually respect the data manifold.

Cut the noise: xai-cola slashes redundant feature changes in counterfactual explanations by up to 50%, making them more interpretable and actionable.