Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox
Picture for Xingjun Ma

Xingjun Ma

Exploring Architectural Ingredients of Adversarially Robust Deep Neural Networks


Oct 13, 2021
Hanxun Huang, Yisen Wang, Sarah Monazam Erfani, Quanquan Gu, James Bailey, Xingjun Ma

Figure 1 for Exploring Architectural Ingredients of Adversarially Robust Deep Neural Networks
Figure 2 for Exploring Architectural Ingredients of Adversarially Robust Deep Neural Networks
Figure 3 for Exploring Architectural Ingredients of Adversarially Robust Deep Neural Networks
Figure 4 for Exploring Architectural Ingredients of Adversarially Robust Deep Neural Networks
* NeurIPS 2021 
 * NeurIPS 2021 
  Access Paper or Ask Questions

Revisiting Adversarial Robustness Distillation: Robust Soft Labels Make Student Better


Aug 18, 2021
Bojia Zi, Shihao Zhao, Xingjun Ma, Yu-Gang Jiang

Figure 1 for Revisiting Adversarial Robustness Distillation: Robust Soft Labels Make Student Better
Figure 2 for Revisiting Adversarial Robustness Distillation: Robust Soft Labels Make Student Better
Figure 3 for Revisiting Adversarial Robustness Distillation: Robust Soft Labels Make Student Better
Figure 4 for Revisiting Adversarial Robustness Distillation: Robust Soft Labels Make Student Better

  Access Paper or Ask Questions

Noise Doesn't Lie: Towards Universal Detection of Deep Inpainting


Jun 03, 2021
Ang Li, Qiuhong Ke, Xingjun Ma, Haiqin Weng, Zhiyuan Zong, Feng Xue, Rui Zhang

Figure 1 for Noise Doesn't Lie: Towards Universal Detection of Deep Inpainting
Figure 2 for Noise Doesn't Lie: Towards Universal Detection of Deep Inpainting
Figure 3 for Noise Doesn't Lie: Towards Universal Detection of Deep Inpainting
Figure 4 for Noise Doesn't Lie: Towards Universal Detection of Deep Inpainting
* Accepted by IJCAI 2021 
  Access Paper or Ask Questions

Dual Head Adversarial Training


Apr 22, 2021
Yujing Jiang, Xingjun Ma, Sarah Monazam Erfani, James Bailey

Figure 1 for Dual Head Adversarial Training
Figure 2 for Dual Head Adversarial Training
Figure 3 for Dual Head Adversarial Training
Figure 4 for Dual Head Adversarial Training

  Access Paper or Ask Questions

Improving Adversarial Robustness via Channel-wise Activation Suppressing


Mar 11, 2021
Yang Bai, Yuyuan Zeng, Yong Jiang, Shu-Tao Xia, Xingjun Ma, Yisen Wang

Figure 1 for Improving Adversarial Robustness via Channel-wise Activation Suppressing
Figure 2 for Improving Adversarial Robustness via Channel-wise Activation Suppressing
Figure 3 for Improving Adversarial Robustness via Channel-wise Activation Suppressing
Figure 4 for Improving Adversarial Robustness via Channel-wise Activation Suppressing
* ICLR2021 accepted paper 
  Access Paper or Ask Questions

Multi-class Classification Based Anomaly Detection of Insider Activities


Feb 15, 2021
R G Gayathri, Atul Sajjanhar, Yong Xiang, Xingjun Ma

Figure 1 for Multi-class Classification Based Anomaly Detection of Insider Activities
Figure 2 for Multi-class Classification Based Anomaly Detection of Insider Activities
Figure 3 for Multi-class Classification Based Anomaly Detection of Insider Activities
Figure 4 for Multi-class Classification Based Anomaly Detection of Insider Activities
* 8 pages, 5 figures 
  Access Paper or Ask Questions

What Do Deep Nets Learn? Class-wise Patterns Revealed in the Input Space


Feb 06, 2021
Shihao Zhao, Xingjun Ma, Yisen Wang, James Bailey, Bo Li, Yu-Gang Jiang

Figure 1 for What Do Deep Nets Learn? Class-wise Patterns Revealed in the Input Space
Figure 2 for What Do Deep Nets Learn? Class-wise Patterns Revealed in the Input Space
Figure 3 for What Do Deep Nets Learn? Class-wise Patterns Revealed in the Input Space
Figure 4 for What Do Deep Nets Learn? Class-wise Patterns Revealed in the Input Space

  Access Paper or Ask Questions

Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networks


Jan 27, 2021
Yige Li, Xixiang Lyu, Nodens Koren, Lingjuan Lyu, Bo Li, Xingjun Ma

Figure 1 for Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networks
Figure 2 for Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networks
Figure 3 for Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networks
Figure 4 for Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networks
* 19 pages, 14 figures, ICLR 2021 
  Access Paper or Ask Questions

Adversarial Interaction Attack: Fooling AI to Misinterpret Human Intentions


Jan 17, 2021
Nodens Koren, Qiuhong Ke, Yisen Wang, James Bailey, Xingjun Ma

Figure 1 for Adversarial Interaction Attack: Fooling AI to Misinterpret Human Intentions
Figure 2 for Adversarial Interaction Attack: Fooling AI to Misinterpret Human Intentions
Figure 3 for Adversarial Interaction Attack: Fooling AI to Misinterpret Human Intentions
Figure 4 for Adversarial Interaction Attack: Fooling AI to Misinterpret Human Intentions
* Preprint 
  Access Paper or Ask Questions

Unlearnable Examples: Making Personal Data Unexploitable


Jan 13, 2021
Hanxun Huang, Xingjun Ma, Sarah Monazam Erfani, James Bailey, Yisen Wang

Figure 1 for Unlearnable Examples: Making Personal Data Unexploitable
Figure 2 for Unlearnable Examples: Making Personal Data Unexploitable
Figure 3 for Unlearnable Examples: Making Personal Data Unexploitable
Figure 4 for Unlearnable Examples: Making Personal Data Unexploitable
* ICLR2021 Spotlight 
  Access Paper or Ask Questions

WildDeepfake: A Challenging Real-World Dataset for Deepfake Detection


Jan 05, 2021
Bojia Zi, Minghao Chang, Jingjing Chen, Xingjun Ma, Yu-Gang Jiang

Figure 1 for WildDeepfake: A Challenging Real-World Dataset for Deepfake Detection
Figure 2 for WildDeepfake: A Challenging Real-World Dataset for Deepfake Detection
Figure 3 for WildDeepfake: A Challenging Real-World Dataset for Deepfake Detection
Figure 4 for WildDeepfake: A Challenging Real-World Dataset for Deepfake Detection

  Access Paper or Ask Questions

Neural Architecture Search via Combinatorial Multi-Armed Bandit


Jan 01, 2021
Hanxun Huang, Xingjun Ma, Sarah M. Erfani, James Bailey

Figure 1 for Neural Architecture Search via Combinatorial Multi-Armed Bandit
Figure 2 for Neural Architecture Search via Combinatorial Multi-Armed Bandit
Figure 3 for Neural Architecture Search via Combinatorial Multi-Armed Bandit
Figure 4 for Neural Architecture Search via Combinatorial Multi-Armed Bandit
* 10 pages, 7 figures 
  Access Paper or Ask Questions

Privacy and Robustness in Federated Learning: Attacks and Defenses


Dec 07, 2020
Lingjuan Lyu, Han Yu, Xingjun Ma, Lichao Sun, Jun Zhao, Qiang Yang, Philip S. Yu

Figure 1 for Privacy and Robustness in Federated Learning: Attacks and Defenses
Figure 2 for Privacy and Robustness in Federated Learning: Attacks and Defenses
Figure 3 for Privacy and Robustness in Federated Learning: Attacks and Defenses
Figure 4 for Privacy and Robustness in Federated Learning: Attacks and Defenses
* arXiv admin note: text overlap with arXiv:2003.02133; text overlap with arXiv:1911.11815 by other authors 
  Access Paper or Ask Questions

Short-Term and Long-Term Context Aggregation Network for Video Inpainting


Sep 12, 2020
Ang Li, Shanshan Zhao, Xingjun Ma, Mingming Gong, Jianzhong Qi, Rui Zhang, Dacheng Tao, Ramamohanarao Kotagiri

Figure 1 for Short-Term and Long-Term Context Aggregation Network for Video Inpainting
Figure 2 for Short-Term and Long-Term Context Aggregation Network for Video Inpainting
Figure 3 for Short-Term and Long-Term Context Aggregation Network for Video Inpainting
Figure 4 for Short-Term and Long-Term Context Aggregation Network for Video Inpainting
* Accepted by ECCV 2020 as a spotlight paper 
  Access Paper or Ask Questions

How to Democratise and Protect AI: Fair and Differentially Private Decentralised Deep Learning


Jul 18, 2020
Lingjuan Lyu, Yitong Li, Karthik Nandakumar, Jiangshan Yu, Xingjun Ma

Figure 1 for How to Democratise and Protect AI: Fair and Differentially Private Decentralised Deep Learning
Figure 2 for How to Democratise and Protect AI: Fair and Differentially Private Decentralised Deep Learning
Figure 3 for How to Democratise and Protect AI: Fair and Differentially Private Decentralised Deep Learning
Figure 4 for How to Democratise and Protect AI: Fair and Differentially Private Decentralised Deep Learning
* Accepted for publication in TDSC 
  Access Paper or Ask Questions

Reflection Backdoor: A Natural Backdoor Attack on Deep Neural Networks


Jul 13, 2020
Yunfei Liu, Xingjun Ma, James Bailey, Feng Lu

Figure 1 for Reflection Backdoor: A Natural Backdoor Attack on Deep Neural Networks
Figure 2 for Reflection Backdoor: A Natural Backdoor Attack on Deep Neural Networks
Figure 3 for Reflection Backdoor: A Natural Backdoor Attack on Deep Neural Networks
Figure 4 for Reflection Backdoor: A Natural Backdoor Attack on Deep Neural Networks
* Accepted by ECCV-2020 
  Access Paper or Ask Questions

Imbalanced Gradients: A New Cause of Overestimated Adversarial Robustness


Jun 30, 2020
Linxi Jiang, Xingjun Ma, Zejia Weng, James Bailey, Yu-Gang Jiang

Figure 1 for Imbalanced Gradients: A New Cause of Overestimated Adversarial Robustness
Figure 2 for Imbalanced Gradients: A New Cause of Overestimated Adversarial Robustness
Figure 3 for Imbalanced Gradients: A New Cause of Overestimated Adversarial Robustness
Figure 4 for Imbalanced Gradients: A New Cause of Overestimated Adversarial Robustness
* 17 pages, 7 figues 
  Access Paper or Ask Questions

Normalized Loss Functions for Deep Learning with Noisy Labels


Jun 24, 2020
Xingjun Ma, Hanxun Huang, Yisen Wang, Simone Romano, Sarah Erfani, James Bailey

Figure 1 for Normalized Loss Functions for Deep Learning with Noisy Labels
Figure 2 for Normalized Loss Functions for Deep Learning with Noisy Labels
Figure 3 for Normalized Loss Functions for Deep Learning with Noisy Labels
Figure 4 for Normalized Loss Functions for Deep Learning with Noisy Labels
* Accepted to ICML 2020 
  Access Paper or Ask Questions

Adversarial Camouflage: Hiding Physical-World Attacks with Natural Styles


Mar 08, 2020
Ranjie Duan, Xingjun Ma, Yisen Wang, James Bailey, A. K. Qin, Yun Yang

Figure 1 for Adversarial Camouflage: Hiding Physical-World Attacks with Natural Styles
Figure 2 for Adversarial Camouflage: Hiding Physical-World Attacks with Natural Styles
Figure 3 for Adversarial Camouflage: Hiding Physical-World Attacks with Natural Styles
Figure 4 for Adversarial Camouflage: Hiding Physical-World Attacks with Natural Styles

  Access Paper or Ask Questions

Clean-Label Backdoor Attacks on Video Recognition Models


Mar 06, 2020
Shihao Zhao, Xingjun Ma, Xiang Zheng, James Bailey, Jingjing Chen, Yu-Gang Jiang

Figure 1 for Clean-Label Backdoor Attacks on Video Recognition Models
Figure 2 for Clean-Label Backdoor Attacks on Video Recognition Models
Figure 3 for Clean-Label Backdoor Attacks on Video Recognition Models
Figure 4 for Clean-Label Backdoor Attacks on Video Recognition Models

  Access Paper or Ask Questions

Skip Connections Matter: On the Transferability of Adversarial Examples Generated with ResNets


Feb 14, 2020
Dongxian Wu, Yisen Wang, Shu-Tao Xia, James Bailey, Xingjun Ma

Figure 1 for Skip Connections Matter: On the Transferability of Adversarial Examples Generated with ResNets
Figure 2 for Skip Connections Matter: On the Transferability of Adversarial Examples Generated with ResNets
Figure 3 for Skip Connections Matter: On the Transferability of Adversarial Examples Generated with ResNets
Figure 4 for Skip Connections Matter: On the Transferability of Adversarial Examples Generated with ResNets
* ICLR 2020 conference paper (spotlight) 
  Access Paper or Ask Questions

Symmetric Cross Entropy for Robust Learning with Noisy Labels


Aug 16, 2019
Yisen Wang, Xingjun Ma, Zaiyi Chen, Yuan Luo, Jinfeng Yi, James Bailey

Figure 1 for Symmetric Cross Entropy for Robust Learning with Noisy Labels
Figure 2 for Symmetric Cross Entropy for Robust Learning with Noisy Labels
Figure 3 for Symmetric Cross Entropy for Robust Learning with Noisy Labels
Figure 4 for Symmetric Cross Entropy for Robust Learning with Noisy Labels
* ICCV2019 
  Access Paper or Ask Questions

Generative Image Inpainting with Submanifold Alignment


Aug 01, 2019
Ang Li, Jianzhong Qi, Rui Zhang, Xingjun Ma, Kotagiri Ramamohanarao

Figure 1 for Generative Image Inpainting with Submanifold Alignment
Figure 2 for Generative Image Inpainting with Submanifold Alignment
Figure 3 for Generative Image Inpainting with Submanifold Alignment
Figure 4 for Generative Image Inpainting with Submanifold Alignment
* accepted by IJCAI2019 
  Access Paper or Ask Questions

Understanding Adversarial Attacks on Deep Learning Based Medical Image Analysis Systems


Jul 24, 2019
Xingjun Ma, Yuhao Niu, Lin Gu, Yisen Wang, Yitian Zhao, James Bailey, Feng Lu

Figure 1 for Understanding Adversarial Attacks on Deep Learning Based Medical Image Analysis Systems
Figure 2 for Understanding Adversarial Attacks on Deep Learning Based Medical Image Analysis Systems
Figure 3 for Understanding Adversarial Attacks on Deep Learning Based Medical Image Analysis Systems
Figure 4 for Understanding Adversarial Attacks on Deep Learning Based Medical Image Analysis Systems
* 15 pages, 10 figures 
  Access Paper or Ask Questions

Towards Fair and Decentralized Privacy-Preserving Deep Learning with Blockchain


Jun 04, 2019
Lingjuan Lyu, Jiangshan Yu, Karthik Nandakumar, Yitong Li, Xingjun Ma, Jiong Jin

Figure 1 for Towards Fair and Decentralized Privacy-Preserving Deep Learning with Blockchain
Figure 2 for Towards Fair and Decentralized Privacy-Preserving Deep Learning with Blockchain
Figure 3 for Towards Fair and Decentralized Privacy-Preserving Deep Learning with Blockchain
Figure 4 for Towards Fair and Decentralized Privacy-Preserving Deep Learning with Blockchain
* 13 pages, 5 figures, 6 tables 
  Access Paper or Ask Questions

Quality Evaluation of GANs Using Cross Local Intrinsic Dimensionality


May 02, 2019
Sukarna Barua, Xingjun Ma, Sarah Monazam Erfani, Michael E. Houle, James Bailey

Figure 1 for Quality Evaluation of GANs Using Cross Local Intrinsic Dimensionality
Figure 2 for Quality Evaluation of GANs Using Cross Local Intrinsic Dimensionality
Figure 3 for Quality Evaluation of GANs Using Cross Local Intrinsic Dimensionality
Figure 4 for Quality Evaluation of GANs Using Cross Local Intrinsic Dimensionality
* The first and original version of this paper was submitted to ICLR 2019 conference. Submission link: https://openreview.net/pdf?id=BJgYl205tQ 
  Access Paper or Ask Questions