Towards continuous learning for glioma segmentation with elastic weight consolidation
Image and Video Processing
2019-09-26 v1 Computer Vision and Pattern Recognition
Abstract
When finetuning a convolutional neural network (CNN) on data from a new domain, catastrophic forgetting will reduce performance on the original training data. Elastic Weight Consolidation (EWC) is a recent technique to prevent this, which we evaluated while training and re-training a CNN to segment glioma on two different datasets. The network was trained on the public BraTS dataset and finetuned on an in-house dataset with non-enhancing low-grade glioma. EWC was found to decrease catastrophic forgetting in this case, but was also found to restrict adaptation to the new domain.
Keywords
Cite
@article{arxiv.1909.11479,
title = {Towards continuous learning for glioma segmentation with elastic weight consolidation},
author = {Karin van Garderen and Sebastian van der Voort and Fatih Incekara and Marion Smits and Stefan Klein},
journal= {arXiv preprint arXiv:1909.11479},
year = {2019}
}