Carbon footprint driven deep learning model selection for medical imaging
Publikation: Konferencebidrag › Paper › Forskning › fagfællebedømt
Standard
Carbon footprint driven deep learning model selection for medical imaging. / Selvan, Raghavendra.
2021. 1-3 Paper præsenteret ved MIDL 2021 International Conference on Medical Imaging with Deep Learning, Lübeck, Tyskland.Publikation: Konferencebidrag › Paper › Forskning › fagfællebedømt
Harvard
APA
Vancouver
Author
Bibtex
}
RIS
TY - CONF
T1 - Carbon footprint driven deep learning model selection for medical imaging
AU - Selvan, Raghavendra
PY - 2021
Y1 - 2021
N2 - Selecting task appropriate deep learning models is a resource intensive process; more so when working with large quantities of high dimensional data that are encountered in medical imaging. Model selection procedures that are primarily aimed at improving performance measures such as accuracy could become biased towards resource intensive models. In this work, we propose to inform and drive the model selection procedure using the carbon footprint of training deep learning models as a complementary measure along with other standard performance metrics. We experimentally demonstrate that increasing carbon footprint of large models might not necessarily translate into proportional performance gains, and suggest useful trade-offs to obtain resource efficient models.
AB - Selecting task appropriate deep learning models is a resource intensive process; more so when working with large quantities of high dimensional data that are encountered in medical imaging. Model selection procedures that are primarily aimed at improving performance measures such as accuracy could become biased towards resource intensive models. In this work, we propose to inform and drive the model selection procedure using the carbon footprint of training deep learning models as a complementary measure along with other standard performance metrics. We experimentally demonstrate that increasing carbon footprint of large models might not necessarily translate into proportional performance gains, and suggest useful trade-offs to obtain resource efficient models.
M3 - Paper
SP - 1
EP - 3
T2 - MIDL 2021 International Conference on Medical Imaging with Deep Learning
Y2 - 7 June 2021 through 7 June 2021
ER -
ID: 287760877