notesum.ai

Published at November 4

Benchmarking XAI Explanations with Human-Aligned Evaluations

cs.CV

cs.AI

cs.HC

Released Date: November 4, 2024

Authors: Rémi Kazmierczak, Steve Azzolin¹, Eloïse Berthier², Anna Hedström, Patricia Delhomme³, Nicolas Bousquet⁴, Goran Frehse⁵, Massimiliano Mancini¹, Baptiste Caramiaux², Andrea Passerini¹, Gianni Franchi¹

Aff.: ¹University of Trento; ²TU Berlin, Germany; ³CNRS, INSA Lyon, UCBL, LIRIS; ⁴Sorbonne Universite, CNRS, ISIR; ⁵U2IS, ENSTA Paris, Institut Polytechnique de Paris

Arxiv: http://arxiv.org/abs/2411.02470v1

	ROAD		Black patches		Uniform noise		Gaussian noise
	PCC	SCC	PCC	SCC	PCC	SCC	PCC	SCC
Q1	0.01 (0.62)	-0.04 (0.12)	0.06 (0.02)	0.05 (0.04)	0.07 (0.01)	0.03 (0.24)	0.07 (0.02)	0.02 (0.38)
Q2	-0.01 (0.88)	-0.03 (0.20)	0.04 (0.19)	0.04 (0.17)	0.06 (0.04)	0.02 (0.45)	0.05 (0.08)	0.01 ( 0.63)
Q3	0.03 (0.33)	-0.03 (0.34)	0.04 (0.10)	0.04 (0.16)	0.08 (0.01)	0.04 (0.13)	0.06 (0.02)	0.03 (0.30)
Q4	0.03 (0.25)	-0.04 (0.17)	0.04 (0.19)	0.03 (0.33)	0.08 (0.01)	0.03 (0.21)	0.06 (0.02)	0.02 (0.41)
Q5	-0.05 (0.05)	0.02 (0.41)	0.01 (0.93)	0.04 (0.14)	-0.05 (0.07)	-0.01 (0.78)	-0.04 (0.16)	0.01 (0.95)
Q6	-0.02 (0.37)	0.06 (0.03)	-0.13 (1e-5)	-0.08 (0.01)	-0.04 (0.16)	-0.01 (0.68)	-0.04 (0.13)	-0.01 (0.79)