IEICE TRANSACTIONS on Information
Robustness of Deep Learning Models in Dermatological Evaluation: A Critical Assessment
Summary :
Our paper attempts to critically assess the robustness of deep learning methods in dermatological evaluation. Although deep learning is being increasingly sought as a means to improve dermatological diagnostics, the performance of models and methods have been rarely investigated beyond studies done under ideal settings. We aim to look beyond results obtained on curated and ideal data corpus, by investigating resilience and performance on user-submitted data. Assessing via few imitated conditions, we have found the overall accuracy to drop and individual predictions change significantly in many cases despite of robust training.
- Publication
- IEICE TRANSACTIONS on Information Vol.E104-D No.3 pp.419-429
- Publication Date
- 2021/03/01
- Publicized
- 2020/12/22
- Online ISSN
- 1745-1361
- DOI
- 10.1587/transinf.2020EDP7133
- Type of Manuscript
- PAPER
- Category
- Artificial Intelligence, Data Mining
Authors
Sourav MISHRA
University of Tokyo
Subhajit CHAUDHURY
University of Tokyo
Hideaki IMAIZUMI
exMedio Inc.
Toshihiko YAMASAKI
University of Tokyo
Keyword
Latest Issue
Copyrights notice of machine-translated contents
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Cite this
Copy
Sourav MISHRA, Subhajit CHAUDHURY, Hideaki IMAIZUMI, Toshihiko YAMASAKI, "Robustness of Deep Learning Models in Dermatological Evaluation: A Critical Assessment" in IEICE TRANSACTIONS on Information,
vol. E104-D, no. 3, pp. 419-429, March 2021, doi: 10.1587/transinf.2020EDP7133.
Abstract: Our paper attempts to critically assess the robustness of deep learning methods in dermatological evaluation. Although deep learning is being increasingly sought as a means to improve dermatological diagnostics, the performance of models and methods have been rarely investigated beyond studies done under ideal settings. We aim to look beyond results obtained on curated and ideal data corpus, by investigating resilience and performance on user-submitted data. Assessing via few imitated conditions, we have found the overall accuracy to drop and individual predictions change significantly in many cases despite of robust training.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2020EDP7133/_p
Copy
@ARTICLE{e104-d_3_419,
author={Sourav MISHRA, Subhajit CHAUDHURY, Hideaki IMAIZUMI, Toshihiko YAMASAKI, },
journal={IEICE TRANSACTIONS on Information},
title={Robustness of Deep Learning Models in Dermatological Evaluation: A Critical Assessment},
year={2021},
volume={E104-D},
number={3},
pages={419-429},
abstract={Our paper attempts to critically assess the robustness of deep learning methods in dermatological evaluation. Although deep learning is being increasingly sought as a means to improve dermatological diagnostics, the performance of models and methods have been rarely investigated beyond studies done under ideal settings. We aim to look beyond results obtained on curated and ideal data corpus, by investigating resilience and performance on user-submitted data. Assessing via few imitated conditions, we have found the overall accuracy to drop and individual predictions change significantly in many cases despite of robust training.},
keywords={},
doi={10.1587/transinf.2020EDP7133},
ISSN={1745-1361},
month={March},}
Copy
TY - JOUR
TI - Robustness of Deep Learning Models in Dermatological Evaluation: A Critical Assessment
T2 - IEICE TRANSACTIONS on Information
SP - 419
EP - 429
AU - Sourav MISHRA
AU - Subhajit CHAUDHURY
AU - Hideaki IMAIZUMI
AU - Toshihiko YAMASAKI
PY - 2021
DO - 10.1587/transinf.2020EDP7133
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E104-D
IS - 3
JA - IEICE TRANSACTIONS on Information
Y1 - March 2021
AB - Our paper attempts to critically assess the robustness of deep learning methods in dermatological evaluation. Although deep learning is being increasingly sought as a means to improve dermatological diagnostics, the performance of models and methods have been rarely investigated beyond studies done under ideal settings. We aim to look beyond results obtained on curated and ideal data corpus, by investigating resilience and performance on user-submitted data. Assessing via few imitated conditions, we have found the overall accuracy to drop and individual predictions change significantly in many cases despite of robust training.
ER -
English
