Timezone: »

Exploring Model Dynamics for Accumulative Poisoning Discovery
Jianing Zhu · Xiawei Guo · Jiangchao Yao · Chao Du · LI He · Shuo Yuan · Tongliang Liu · Liang Wang · Bo Han

Wed Jul 26 02:00 PM -- 03:30 PM (PDT) @ Exhibit Hall 1 #630

Adversarial poisoning attacks pose huge threats to various machine learning applications. Especially, the recent accumulative poisoning attacks show that it is possible to achieve irreparable harm on models via a sequence of imperceptible attacks followed by a trigger batch. Due to the limited data-level discrepancy in real-time data streaming, current defensive methods are indiscriminate in handling the poison and clean samples. In this paper, we dive into the perspective of model dynamics and propose a novel information measure, namely, Memorization Discrepancy, to explore the defense via the model-level information. By implicitly transferring the changes in the data manipulation to that in the model outputs, Memorization Discrepancy can discover the imperceptible poison samples based on their distinct dynamics from the clean samples. We thoroughly explore its properties and propose Discrepancy-aware Sample Correction (DSC) to defend against accumulative poisoning attacks. Extensive experiments comprehensively characterized Memorization Discrepancy and verified its effectiveness. The code is publicly available at: https://github.com/tmlr-group/Memorization-Discrepancy.

Author Information

Jianing Zhu (Hong Kong Baptist University)
Xiawei Guo (Department of Computer Science and Engineering, Hong Kong University of Science and Technology)
Jiangchao Yao (Cooperative Medianet Innovation Center, Shang hai Jiao Tong University)
Chao Du (Sea AI Lab)
LI He (Tsinghua University, Tsinghua University)
Shuo Yuan (Alibaba Group)
Tongliang Liu (The University of Sydney)
Liang Wang (Alibaba Group)

More from the Same Authors