Home Publications Research Grants Inventions & Patents Awards Additional Research Activities Faculties & Institutions Research Areas

The Effect of the Loss on Generalization: Empirical Study on Synthetic Lung Nodule Data

Baltatzis V, Le Folgoc L, Ellis S, Manzanera OEM, Bintsi KM, Nair A, Desai S, Glocker B, Schnabel JA (2021)

---

Publication Type: Conference contribution

Publication year: 2021

Journal

Publisher: Springer Science and Business Media Deutschland GmbH

Book Volume: 12929 LNCS

Pages Range: 56-64

Conference Proceedings Title: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

Event location: Strasbourg, FRA

ISBN: 9783030874438

DOI: 10.1007/978-3-030-87444-5_6

Abstract

Convolutional Neural Networks (CNNs) are widely used for image classification in a variety of fields, including medical imaging. While most studies deploy cross-entropy as the loss function in such tasks, a growing number of approaches have turned to a family of contrastive learning-based losses. Even though performance metrics such as accuracy, sensitivity and specificity are regularly used for the evaluation of CNN classifiers, the features that these classifiers actually learn are rarely identified and their effect on the classification performance on out-of-distribution test samples is insufficiently explored. In this paper, motivated by the real-world task of lung nodule classification, we investigate the features that a CNN learns when trained and tested on different distributions of a synthetic dataset with controlled modes of variation. We show that different loss functions lead to different features being learned and consequently affect the generalization ability of the classifier on unseen data. This study provides some important insights into the design of deep learning solutions for medical imaging tasks.

Involved external institutions

How to cite

APA:

Baltatzis, V., Le Folgoc, L., Ellis, S., Manzanera, O.E.M., Bintsi, K.-M., Nair, A.,... Schnabel, J.A. (2021). The Effect of the Loss on Generalization: Empirical Study on Synthetic Lung Nodule Data. In Mauricio Reyes, Pedro Henriques Abreu, Jaime Cardoso, Mustafa Hajij, Ghada Zamzmi, Paul Rahul, Lokendra Thakur (Eds.), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (pp. 56-64). Strasbourg, FRA: Springer Science and Business Media Deutschland GmbH.

MLA:

Baltatzis, Vasileios, et al. "The Effect of the Loss on Generalization: Empirical Study on Synthetic Lung Nodule Data." Proceedings of the 4th International Workshop on Interpretability of Machine Intelligence in Medical Image Computing, iMIMIC 2020 and 1st International Workshop on Topological Data Analysis and Its Applications for Medical Data, TDA4MedicalData 2021 held in conjunction with 24th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2021, Strasbourg, FRA Ed. Mauricio Reyes, Pedro Henriques Abreu, Jaime Cardoso, Mustafa Hajij, Ghada Zamzmi, Paul Rahul, Lokendra Thakur, Springer Science and Business Media Deutschland GmbH, 2021. 56-64.

BibTeX: Download