Session 7D: ML Security
Authors, creators, and presenters: Rui Wen (CISPA Helmholtz Center for Information Security), Michael Backes (CISPA Helmholtz Center for Information Security), Yang Zhang (CISPA Helmholtz Center for Information Security)
paper
Understanding the importance of data in machine learning attacks: Does valuable data cause more harm?
Machine learning is revolutionizing many fields, playing a key role in driving progress and enabling data-centric processes. The importance of data in training a model and shaping its performance cannot be overstated. Recent studies have highlighted the heterogeneous influence of individual data samples, especially the presence of valuable data that significantly contributes to the usefulness and effectiveness of machine learning models. However, important questions remain unanswered. Are these valuable data samples more vulnerable to machine learning attacks? In this study, we investigate the relationship between data criticality and machine learning attacks by analyzing five different attack types. Our findings reveal some notable insights. For example, we have observed that high-severity data samples have increased vulnerabilities to certain attacks, such as membership inference and model theft. These findings also have practical implications, encouraging researchers to design more efficient attacks. By analyzing the association between membership inference vulnerabilities and data importance, we show that introducing sample-specific criteria can integrate sample characteristics into membership metrics, thereby improving membership inference performance. These findings highlight the urgent need for innovative defense mechanisms that strike a balance between maximizing utility and protecting valuable data from potential misuse.
About NDSS
The Network and Distributed Systems Security Symposium (NDSS) facilitates the exchange of information between network and distributed systems security researchers and practitioners. The intended audience includes anyone interested in the practical aspects of network and distributed systems security, with an emphasis on practical system design and implementation. A primary goal is to encourage and enable the Internet community to adapt, adopt, and advance available security technologies.
We would like to thank the Network and Distributed Systems Security (NDSS) Symposium creators, authors, and presenters for sharing great content from the NDSS Symposium 2025 conference on their organization's YouTube channel.
Permalink
*** This is a syndicated blog from the Security Bloggers Network brought to you by Infosecurity.US and written by Marc Handelman. Read the original post: https://www.youtube-nocookie.com/embed/BEqhZDdleDs?si=BbDog9eTwm7E5zY3
Source link
