Replay-Oriented Gradient Projection Memory for Continual Learning in Medical Scenarios

Kuang Shu, Heng Li, Jie Cheng, Qinghai Guo, Luziwei Leng, Jianxing Liao, Yan Hu, Jiang Liu

Research output: Chapter in Book/Conference proceedingConference contributionpeer-review

3 Citations (Scopus)

Abstract

Despite the tremendous progress recently achieved by deep learning (DL) in medical image analysis, most DL models only concentrate on single data distribution, which follows the independent and identically distributed (i.i.d) assumption. However, in practice, image data distribution changes with clinical conditions, such as different scanner manufacturers, imaging settings, and statistics regions. Although one can further train the model on new data samples, updating a model with data from an unknown distribution will always result in the model's performance degradation on the learned data, a notorious phenomenon called catastrophic forgetting. Therefore affects the applicability of DL algorithms in continuously changing clinical scenarios. In this study, we have proposed a new method to address the impact of changing distributions in continual learning scenarios and alleviate catastrophic forgetting. A gradient regularization approach is used to suppress forgetting, and a replay-oriented consistency calculation method combined with a subspace weighting strategy is proposed to improve the model plasticity further. The proposed replay-oriented gradient projection memory (RO-GPM) is evaluated on multiple fundus disease diagnosis datasets including a real-world application and a continual learning benchmark. The quantitative and visualization results demonstrate that the proposed RO-GPM achieves superior performance to state-of-the-art algorithms by a large margin.1

Original languageEnglish
Title of host publicationProceedings - 2022 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2022
EditorsDonald Adjeroh, Qi Long, Xinghua Shi, Fei Guo, Xiaohua Hu, Srinivas Aluru, Giri Narasimhan, Jianxin Wang, Mingon Kang, Ananda M. Mondal, Jin Liu
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1724-1729
Number of pages6
ISBN (Electronic)9781665468190
DOIs
Publication statusPublished - 2022
Externally publishedYes
Event2022 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2022 - Las Vegas, United States
Duration: 6 Dec 20228 Dec 2022

Publication series

NameProceedings - 2022 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2022

Conference

Conference2022 IEEE International Conference on Bioinformatics and Biomedicine, BIBM 2022
Country/TerritoryUnited States
CityLas Vegas
Period6/12/228/12/22

Keywords

  • Continual learning
  • catastrophic forgetting
  • gradient regularization
  • multiple fundus disease
  • replay strategy

ASJC Scopus subject areas

  • Psychiatry and Mental health
  • Information Systems and Management
  • Biomedical Engineering
  • Medicine (miscellaneous)
  • Cardiology and Cardiovascular Medicine
  • Health Informatics

Fingerprint

Dive into the research topics of 'Replay-Oriented Gradient Projection Memory for Continual Learning in Medical Scenarios'. Together they form a unique fingerprint.

Cite this