Fine-Grained Poisoning Attack to Local Differential Privacy Protocols for Mean and Variance Estimation
Offered By: USENIX via YouTube
Course Description
Overview
Explore a cutting-edge research presentation on data poisoning attacks against local differential privacy (LDP) protocols. Delve into the proposed fine-grained attack that manipulates mean and variance estimations in LDP systems. Learn about the novel output poisoning attack (OPA) technique, which injects fake data into the output domain of local LDP instances. Examine the security-privacy consistency observed in LDP and gain insights into the evolving threat landscape of data poisoning attacks. Compare the effectiveness of OPA against baseline attacks using real-world datasets. Discover a new defense method for recovering result accuracy from polluted data collections and understand implications for secure LDP design. This 20-minute conference talk from USENIX Security '23 offers valuable knowledge for researchers and practitioners in the fields of privacy-preserving data analysis and cybersecurity.
Syllabus
USENIX Security '23 - Fine-grained Poisoning Attack to Local Differential Privacy Protocols for...
Taught by
USENIX
Related Courses
Natural Language ProcessingColumbia University via Coursera Intro to Algorithms
Udacity Conception et mise en œuvre d'algorithmes.
École Polytechnique via Coursera Paradigms of Computer Programming
Université catholique de Louvain via edX Data Structures and Algorithm Design Part I | 数据结构与算法设计(上)
Tsinghua University via edX