TY - GEN
T1 - Visual Explanation Generation Based on Lambda Attention Branch Networks
AU - Iida, Tsumugi
AU - Komatsu, Takumi
AU - Kaneda, Kanta
AU - Hirakawa, Tsubasa
AU - Yamashita, Takayoshi
AU - Fujiyoshi, Hironobu
AU - Sugiura, Komei
N1 - Funding Information:
Acknowledgement. This work was partially supported by JSPS KAKENHI Grant Number 20H04269 and NEDO.
Publisher Copyright:
© 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.
PY - 2023
Y1 - 2023
N2 - Explanation generation for transformers enhances accountability for their predictions. However, there have been few studies on generating visual explanations for the transformers that use multidimensional context, such as LambdaNetworks. In this paper, we propose the Lambda Attention Branch Networks, which attend to important regions in detail and generate easily interpretable visual explanations. We also propose the Patch Insertion-Deletion score, an extension of the Insertion-Deletion score, as an effective evaluation metric for images with sparse important regions. Experimental results on two public datasets indicate that the proposed method successfully generates visual explanations.
AB - Explanation generation for transformers enhances accountability for their predictions. However, there have been few studies on generating visual explanations for the transformers that use multidimensional context, such as LambdaNetworks. In this paper, we propose the Lambda Attention Branch Networks, which attend to important regions in detail and generate easily interpretable visual explanations. We also propose the Patch Insertion-Deletion score, an extension of the Insertion-Deletion score, as an effective evaluation metric for images with sparse important regions. Experimental results on two public datasets indicate that the proposed method successfully generates visual explanations.
KW - Attention
KW - Lambda networks
KW - Transformer
UR - http://www.scopus.com/inward/record.url?scp=85149631306&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85149631306&partnerID=8YFLogxK
U2 - 10.1007/978-3-031-26284-5_29
DO - 10.1007/978-3-031-26284-5_29
M3 - Conference contribution
AN - SCOPUS:85149631306
SN - 9783031262838
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 475
EP - 490
BT - Computer Vision – ACCV 2022 - 16th Asian Conference on Computer Vision, 2022, Proceedings
A2 - Wang, Lei
A2 - Gall, Juergen
A2 - Chin, Tat-Jun
A2 - Sato, Imari
A2 - Chellappa, Rama
PB - Springer Science and Business Media Deutschland GmbH
T2 - 16th Asian Conference on Computer Vision, ACCV 2022
Y2 - 4 December 2022 through 8 December 2022
ER -