File Download
There are no files associated with this item.
Links for fulltext
(May Require Subscription)
- Publisher Website: 10.1609/aaai.v33i01.33018417
- WOS: WOS:000486572502116
- Find via
Supplementary
-
Citations:
- Web of Science: 0
- Appears in Collections:
Conference Paper: Non-Local Context Encoder: Robust Biomedical Image Segmentation against Adversarial Attack
Title | Non-Local Context Encoder: Robust Biomedical Image Segmentation against Adversarial Attack |
---|---|
Authors | |
Issue Date | 2019 |
Publisher | AAAI Press. The Journal's web site is located at https://aaai.org/Library/AAAI/aaai-library.php |
Citation | The Thirty-Third Association for the Advancement of Artificial Intelligence (AAAI) Conference on Artificial Intelligence (AAAI-19), Honolulu, Hawaii, USA, 27 January– 1 February 2019, v. 33 n. 1, p. 8417-8424 How to Cite? |
Abstract | Recent progress in biomedical image segmentation based on deep convolutional neural networks (CNNs) has drawn much attention. However, its vulnerability towards adversarial samples cannot be overlooked. This paper is the first one that discovers that all the CNN-based state-of-the-art biomedical image segmentation models are sensitive to adversarial perturbations. This limits the deployment of these methods in safety-critical biomedical fields. In this paper, we discover that global spatial dependencies and global contextual information in a biomedical image can be exploited to defend against adversarial attacks. To this end, non-local context encoder (NLCE) is proposed to model short- and long-range spatial dependencies and encode global contexts for strengthening feature activations by channel-wise attention. The NLCE modules enhance the robustness and accuracy of the non-local context encoding network (NLCEN), which learns robust enhanced pyramid feature representations with NLCE modules, and then integrates the information across different levels. Experiments on both lung and skin lesion segmentation datasets have demonstrated that NLCEN outperforms any other state-of-the-art biomedical image segmentation methods against adversarial attacks. In addition, NLCE modules can be applied to improve the robustness of other CNN-based biomedical image segmentation methods. |
Description | Oral presentation: Tech Session 7: Vision (General) 2 - no. 1750 |
Persistent Identifier | http://hdl.handle.net/10722/271321 |
ISSN | |
ISI Accession Number ID |
DC Field | Value | Language |
---|---|---|
dc.contributor.author | He, X | - |
dc.contributor.author | Yang, S | - |
dc.contributor.author | Li, G | - |
dc.contributor.author | Li, H | - |
dc.contributor.author | Chang, H | - |
dc.contributor.author | Yu, Y | - |
dc.date.accessioned | 2019-06-24T01:07:35Z | - |
dc.date.available | 2019-06-24T01:07:35Z | - |
dc.date.issued | 2019 | - |
dc.identifier.citation | The Thirty-Third Association for the Advancement of Artificial Intelligence (AAAI) Conference on Artificial Intelligence (AAAI-19), Honolulu, Hawaii, USA, 27 January– 1 February 2019, v. 33 n. 1, p. 8417-8424 | - |
dc.identifier.issn | 2159-5399 | - |
dc.identifier.uri | http://hdl.handle.net/10722/271321 | - |
dc.description | Oral presentation: Tech Session 7: Vision (General) 2 - no. 1750 | - |
dc.description.abstract | Recent progress in biomedical image segmentation based on deep convolutional neural networks (CNNs) has drawn much attention. However, its vulnerability towards adversarial samples cannot be overlooked. This paper is the first one that discovers that all the CNN-based state-of-the-art biomedical image segmentation models are sensitive to adversarial perturbations. This limits the deployment of these methods in safety-critical biomedical fields. In this paper, we discover that global spatial dependencies and global contextual information in a biomedical image can be exploited to defend against adversarial attacks. To this end, non-local context encoder (NLCE) is proposed to model short- and long-range spatial dependencies and encode global contexts for strengthening feature activations by channel-wise attention. The NLCE modules enhance the robustness and accuracy of the non-local context encoding network (NLCEN), which learns robust enhanced pyramid feature representations with NLCE modules, and then integrates the information across different levels. Experiments on both lung and skin lesion segmentation datasets have demonstrated that NLCEN outperforms any other state-of-the-art biomedical image segmentation methods against adversarial attacks. In addition, NLCE modules can be applied to improve the robustness of other CNN-based biomedical image segmentation methods. | - |
dc.language | eng | - |
dc.publisher | AAAI Press. The Journal's web site is located at https://aaai.org/Library/AAAI/aaai-library.php | - |
dc.relation.ispartof | Proceedings of the AAAI Conference on Artificial Intelligence | - |
dc.title | Non-Local Context Encoder: Robust Biomedical Image Segmentation against Adversarial Attack | - |
dc.type | Conference_Paper | - |
dc.identifier.email | Yu, Y: yzyu@cs.hku.hk | - |
dc.identifier.authority | Yu, Y=rp01415 | - |
dc.identifier.doi | 10.1609/aaai.v33i01.33018417 | - |
dc.identifier.hkuros | 297945 | - |
dc.identifier.volume | 33 | - |
dc.identifier.issue | 1 | - |
dc.identifier.spage | 8417 | - |
dc.identifier.epage | 8424 | - |
dc.identifier.isi | WOS:000486572502116 | - |
dc.publisher.place | United States | - |
dc.identifier.issnl | 2159-5399 | - |