Shortcomings of Interpretability Taxonomies for Deep Neural Networks
Research output: Contribution to journal › Conference article › Research › peer-review
Standard
Shortcomings of Interpretability Taxonomies for Deep Neural Networks. / Søgaard, Anders.
In: CEUR Workshop Proceedings, Vol. 3318, 2022.Research output: Contribution to journal › Conference article › Research › peer-review
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - GEN
T1 - Shortcomings of Interpretability Taxonomies for Deep Neural Networks
AU - Søgaard, Anders
N1 - Publisher Copyright: © 2022 Copyright for this paper by its authors.
PY - 2022
Y1 - 2022
N2 - Taxonomies are vehicles for thinking about what’s possible, for identifying unconsidered options, as well as for establishing formal relations between entities. We identify several shortcomings in 10 existing taxonomies for interpretability methods for explainable artificial intelligence (XAI), focusing on methods for deep neural networks. The shortcomings include redundancies, incompleteness, and inconsistencies. We design a new taxonomy based on two orthogonal dimensions and show how it can be used to derive results about entire classes of interpretability methods for deep neural networks.
AB - Taxonomies are vehicles for thinking about what’s possible, for identifying unconsidered options, as well as for establishing formal relations between entities. We identify several shortcomings in 10 existing taxonomies for interpretability methods for explainable artificial intelligence (XAI), focusing on methods for deep neural networks. The shortcomings include redundancies, incompleteness, and inconsistencies. We design a new taxonomy based on two orthogonal dimensions and show how it can be used to derive results about entire classes of interpretability methods for deep neural networks.
KW - interpretability, taxonomy
M3 - Conference article
AN - SCOPUS:85146255048
VL - 3318
JO - CEUR Workshop Proceedings
JF - CEUR Workshop Proceedings
SN - 1613-0073
T2 - 2022 International Conference on Information and Knowledge Management Workshops, CIKM-WS 2022
Y2 - 17 October 2022 through 21 October 2022
ER -
ID: 336294291