Mahdi Dhaini
@mahdidh.bsky.social
PhD candidate in Trustworthy and Responsible NLP @ Technical University of Munich (TUM)
In this paper, we show that widely used post-hoc feature attribution methods exhibit significant gender disparity with respect to their faithfulness, robustness, and complexity.
This work was done with Ege Erdogan, @nfel.bsky.social , and Gjergji Kasneci.
This work was done with Ege Erdogan, @nfel.bsky.social , and Gjergji Kasneci.
June 26, 2025 at 5:25 AM
In this paper, we show that widely used post-hoc feature attribution methods exhibit significant gender disparity with respect to their faithfulness, robustness, and complexity.
This work was done with Ege Erdogan, @nfel.bsky.social , and Gjergji Kasneci.
This work was done with Ege Erdogan, @nfel.bsky.social , and Gjergji Kasneci.