2021
DOI: 10.48550/arxiv.2101.12602
|View full text |Cite
Preprint
|
Sign up to set email alerts
|

On the differential privacy of dynamic location obfuscation with personalized error bounds

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
1

Citation Types

0
0
0

Year Published

2023
2023
2023
2023

Publication Types

Select...
1

Relationship

0
1

Authors

Journals

citations
Cited by 1 publication
(1 citation statement)
references
References 3 publications
0
0
0
Order By: Relevance
“…Existing works [5], [37] proposed optimal mechanisms for GeoI to consider utility and privacy with prior knowledge using Shokri's notion [25], which we call quality loss (Q loss ) and adversarial error (AE) while guaranteeing ε-GeoI. Concretely, Yu et al [37] proposed a privacy guarantee based on AE in addition to ε-GeoI, but its method does not strictly guarantee ε-GeoI as proved by Shun et al [27]; Bordenabe et al [5] optimized Q loss while guaranteeing GeoI using δ-spanner to improve the runtime of optimization. We can apply δ-spanner to GeoGI, but the method requires that the number of possible locations input to the mechanism is small (e.g., < 100), which causes utility and privacy loss for a user at a location outside the possible input locations as described in Sect.…”
Section: Introductionmentioning
confidence: 99%
“…Existing works [5], [37] proposed optimal mechanisms for GeoI to consider utility and privacy with prior knowledge using Shokri's notion [25], which we call quality loss (Q loss ) and adversarial error (AE) while guaranteeing ε-GeoI. Concretely, Yu et al [37] proposed a privacy guarantee based on AE in addition to ε-GeoI, but its method does not strictly guarantee ε-GeoI as proved by Shun et al [27]; Bordenabe et al [5] optimized Q loss while guaranteeing GeoI using δ-spanner to improve the runtime of optimization. We can apply δ-spanner to GeoGI, but the method requires that the number of possible locations input to the mechanism is small (e.g., < 100), which causes utility and privacy loss for a user at a location outside the possible input locations as described in Sect.…”
Section: Introductionmentioning
confidence: 99%