• English
    • Tiếng Việt
  • Tiếng Việt 
    • English
    • Tiếng Việt
  • Đăng nhập
View Item 
  •   Trang chủ
  • The College of Engineering and Computer Science
  • Nguyen Do Trung Chanh, PhD
  • View Item
  •   Trang chủ
  • The College of Engineering and Computer Science
  • Nguyen Do Trung Chanh, PhD
  • View Item
JavaScript is disabled for your browser. Some features of this site may not work without it.

Diffeomorphism matching for fast unsupervised pretraining on radiographs

Thumbnail
Xem/Mở
nguyendotrungchanhh.pdf (1.390Mb)
Năm xuất bản
2020
Tác giả
Huynh, M. Thanh
Truong, Q. H. Steven
Nguyen, D. T. Chanh
Ta, Duc Huy
Hoang, Cao Huyen
Bui, Trung
Metadata
Hiển thị đầy đủ biểu ghi
Tóm tắt
Unsupervised pretraining is an approach that leverages a large unlabeled data pool to learn data features. However, it requires billion-scale datasets and a month-long training time to surpass its supervised counterpart on fine-tuning in many computer vision tasks. In this study, we propose a novel method, Diffeomorphism Matching (DM), to overcome those challenges. The proposed method combines self-supervised learning and knowledge distillation to equivalently map the feature space of a student model to that of a big pretrained teacher model. On the Chest X-ray dataset, our method alleviates the need to acquire billions of radiographs and substantially reduces pretraining time by 95%. In addition, our pretrained model outperforms other pretrained models by at least 4.2% in F1 score on the CheXpert dataset and 0.7% in Dice score on the SIIM Pneumothorax dataset. Code and pretrained model are available at https://github.com/jokingbear/DM.git.
Định danh
https://vinspace.edu.vn/handle/VIN/618
Collections
  • Nguyen Do Trung Chanh, PhD [11]

Liên hệ | Gửi phản hồi
 

 

Duyệt theo

Toàn bộ thư việnĐơn vị và Bộ sưu tậpNăm xuất bảnTác giảNhan đềChủ đềTrong Bộ sưu tậpNăm xuất bảnTác giảNhan đềChủ đề

Tài khoản

Đăng nhậpĐăng ký

Liên hệ | Gửi phản hồi