MLSP-L29.5
Improving Maximum Margin Backdoor Detection by Class Subspace Decorrelation
Guangmingmei Yang, George Kesidis, David Miller, Pennsylvania State University, United States of America
Session:
MLSP-L29: Adversarial ML – Adversarial ML Oral
Track:
Machine Learning for Signal Processing [ML]
Location:
Room 112
Presentation Time:
Fri, 8 May, 10:20 - 10:40
Presentation
Discussion
Resources
No resources available.
Session MLSP-L29
MLSP-L29.1: DADAGAN: AN IMAGE SUPER-RESOLUTION NETWORK WITH PIXEL-WISE ESTIMATION OF DEGRADATION DEGREES
Xinyue Zhan, Jiaxiong Liu, Man Zhao, Jiaran Qiu, Jun Zhou, Southwest University, China
MLSP-L29.2: Soft Super-Pixel Partitioning for Certified Adversarial Robustness
Hossein Goli, PhD Student, Iran (Islamic Republic of); Farzan Farnia, Assistant Professor, Hong Kong
MLSP-L29.3: DCINJECT: PERSISTENT BACKDOOR ATTACKS VIA FREQUENCY MANIPULATION IN PERSONAL FEDERATED LEARNING
Nahom Birhan, Old Dominion University, United States of America; Daniel Wesego, university of Illinois Chicago, United States of America; Dereje Shenkut, Carnegie Mellon University, United States of America; Frank Liu, Daniel Takabi, Old Dominion University, United States of America
MLSP-L29.4: BDBR: TOWARDS RESISTANT BACKDOOR DEFENSE VIA BOUNDARY RECONSTRUCTION
Hengrui Yan, Xinran Zheng, Tsinghua University, China; Shuo Yang, The University of Hong Kong, Hong Kong; Xingjun Wang, Tsinghua University, China
MLSP-L29.5: Improving Maximum Margin Backdoor Detection by Class Subspace Decorrelation
Guangmingmei Yang, George Kesidis, David Miller, Pennsylvania State University, United States of America
MLSP-L29.6: DCSF: ENHANCING CERTIFIED ROBUSTNESS VIA DYNAMIC COST-SENSITIVE AND SELF-SUPERVISION FRAMEWORK
Xuan Tang, Yanchun Li, Xiangtan University, China; Long Huang, Northwestern Polytechnical University, China; Li Zeng, Xingxia Dai, Xiangtan University, China
Contacts