CONFERENCE PROCEEDINGS
Advanced Search
Home > Proceedings > Volume 13513 > Article
Open Access Paper
15 January 2025 Research on two-party PSI protocol in the unbalanced scenario
Wenjing Nie, Yanshan Nie
Author Affiliations +
Proceedings Volume 13513, The International Conference Optoelectronic Information and Optical Engineering (OIOE2024); 135132J (2025) https://doi.org/10.1117/12.3045721
Event: The International Conference Optoelectronic Information and Optical Engineering (OIOE2024), 2024, Wuhan, China
Abstract
At present, PSI is an indispensable intermediate link in the learning process of vertical federation. When longitudinal federation learning online prediction, PSI running time directly affects online prediction time. To improve the online prediction speed of vertical federation learning, the existing PSI has been improved to reduce the number of matches between two datasets by adding slicing filters, reducing the amount of matching data, and dividing the data sets into buckets. Experiments show that When the data volume of the two parties is unbalanced, and the data volume of the demand side remains unchanged, the more obvious the gap between the running time of the SOPRF-PSI protocol and the other two protocols is as the data volume of the provider increases, the greater the performance advantage of the SOPRF-PSI protocol.
(2025) Published by SPIE. Downloading of the abstract is permitted for personal use only.
Citation Download Citation
Wenjing Nie and Yanshan Nie "Research on two-party PSI protocol in the unbalanced scenario", Proc. SPIE 13513, The International Conference Optoelectronic Information and Optical Engineering (OIOE2024), 135132J (15 January 2025); https://doi.org/10.1117/12.3045721
PROCEEDINGS
 6 PAGES
DOWNLOAD PAPER SAVE TO MY LIBRARY
GET CITATION
Advertisement
Advertisement
RIGHTS & PERMISSIONS
Get copyright permission  Get copyright permission on Copyright Marketplace
KEYWORDS
Tunable filters
Binary data
Data transmission
Data modeling
Computer security
Online learning
Modeling
RELATED CONTENT
Subscribe to Digital Library
Receive Erratum Email Alert
Back to Top