Ching Lam CHOIDebiased path attribution by perturbing in the unlearning direction of steepest ascent.
ICLR 2025Oral
UNI: Unlearning-based Neural Interpretations
Related work
Preprint 2023
On the Generalization of Gradient-based Neural Network Interpretations
Gradient-based XAI fails to generalise from train to test data: regularisation crucially helps.