Login (DCU Staff Only)
Login (DCU Staff Only)

DORAS | DCU Research Repository

Explore open access research and scholarly works from DCU

Advanced Search

Integrating feature attribution methods into the loss function of deep learning classifiers

Callanan, James, Garcia-Cabrera, Carles orcid logoORCID: 0000-0001-8139-9647, Belton, Niamh orcid logoORCID: 0000-0003-4949-4745, Roshchupkin, Gennady and Curran, Kathleen orcid logoORCID: 0000-0003-0095-9337 (2022) Integrating feature attribution methods into the loss function of deep learning classifiers. In: 24th Irish Machine Vision and Image Processing Conference, 31 Aug - 2 Sept 2022, Belfast. ISBN 978-0-9934207-7-1

Feature attribution methods are typically used post-training to judge if a deep learning classifier is using meaningful concepts in an input image when making classifications. In this study, we propose using feature attribution methods to give a classifier automated feedback throughout the training process via a novel loss function. We call such a loss function, a heatmap loss function. Heatmap loss functions enable us to incentivize a model to rely on relevant sections of the input image when making classifications. Two groups of models were trained, one group with a heatmap loss function and the other using categorical cross entropy (CCE). Models trained with the heatmap loss function were capable of achieving equivalent classification accuracies on a test dataset of synthesised cardiac MRIs. Moreover, HiResCAM heatmaps suggest that these models relied to a greater extent on regions of the input image within the heart. A further experiment demonstrated how heatmap loss functions can be used to prevent deep learning classifiers from using non-causal concepts that disproportionately co-occur with certain classes when making classifications. This suggests that heatmap loss functions could be used to prevent models from learning dataset biases by directing where the model should be looking when making classifications.
Item Type:Conference or Workshop Item (Paper)
Event Type:Conference
Uncontrolled Keywords:Loss function; Dataset bias; Grad-CAM; HiResCAM; Deep learning
DCU Faculties and Centres:DCU Faculties and Schools > Faculty of Engineering and Computing > School of Electronic Engineering
Published in: Irish Pattern Recognition & Classification Society Conference Proceedings 2022. . Irish Pattern Recognition & Classification Society (IPRCS). ISBN 978-0-9934207-7-1
Publisher:Irish Pattern Recognition & Classification Society (IPRCS)
Official URL:https://doi.org/10.56541/OMXA8857
Copyright Information:© 2022 The Authors & Publisher.
Funders:Science Foundation Ireland (SFI) Centre for Research Training in Machine Learning (18/CRT/6183).
ID Code:27762
Deposited On:20 Sep 2022 09:08 by Carles Garcia Cabrera . Last Modified 20 Sep 2022 09:41

Full text available as:

[thumbnail of IMVIP2022_paper_1224.pdf]
PDF - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader


Downloads per month over past year

Archive Staff Only: edit this record