TY - GEN
T1 - Importance Driven Continual Learning for Segmentation Across Domains
AU - Özgün, Sinan
AU - Rickmann, Anne Marie
AU - Roy, Abhijit Guha
AU - Wachinger, Christian
N1 - Publisher Copyright:
© 2020, Springer Nature Switzerland AG.
PY - 2020
Y1 - 2020
N2 - The ability of neural networks to continuously learn and adapt to new tasks while retaining prior knowledge is crucial for many applications. However, current neural networks tend to forget previously learned tasks when trained on new ones, i.e., they suffer from Catastrophic Forgetting (CF). The objective of Continual Learning (CL) is to alleviate this problem, which is particularly relevant for medical applications, where it may not be feasible to store and access previously used sensitive patient data. In this work, we propose a Continual Learning approach for brain segmentation, where a single network is consecutively trained on samples from different domains. We build upon an importance driven approach and adapt it for medical image segmentation. Particularly, we introduce a learning rate regularization to prevent the loss of the network’s knowledge. Our results demonstrate that directly restricting the adaptation of important network parameters clearly reduces Catastrophic Forgetting for segmentation across domains. Our code is publicly available on https://github.com/ai-med/MAS-LR.
AB - The ability of neural networks to continuously learn and adapt to new tasks while retaining prior knowledge is crucial for many applications. However, current neural networks tend to forget previously learned tasks when trained on new ones, i.e., they suffer from Catastrophic Forgetting (CF). The objective of Continual Learning (CL) is to alleviate this problem, which is particularly relevant for medical applications, where it may not be feasible to store and access previously used sensitive patient data. In this work, we propose a Continual Learning approach for brain segmentation, where a single network is consecutively trained on samples from different domains. We build upon an importance driven approach and adapt it for medical image segmentation. Particularly, we introduce a learning rate regularization to prevent the loss of the network’s knowledge. Our results demonstrate that directly restricting the adaptation of important network parameters clearly reduces Catastrophic Forgetting for segmentation across domains. Our code is publicly available on https://github.com/ai-med/MAS-LR.
UR - http://www.scopus.com/inward/record.url?scp=85092743820&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-59861-7_43
DO - 10.1007/978-3-030-59861-7_43
M3 - Conference contribution
AN - SCOPUS:85092743820
SN - 9783030598600
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 423
EP - 433
BT - Machine Learning in Medical Imaging - 11th International Workshop, MLMI 2020, Held in Conjunction with MICCAI 2020, Proceedings
A2 - Liu, Mingxia
A2 - Lian, Chunfeng
A2 - Yan, Pingkun
A2 - Cao, Xiaohuan
PB - Springer Science and Business Media Deutschland GmbH
T2 - 11th International Workshop on Machine Learning in Medical Imaging, MLMI 2020, held in conjunction with the 23rd International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2020
Y2 - 4 October 2020 through 4 October 2020
ER -