The Crowd Wisdom for Location Privacy of Crowdsensing Photos: Spear or Shield?
summary
A no-reference location-inference pipeline extracts end-to-end visual features, builds a refined similarity graph, applies spectral clustering, picks representative seeds for crowdsourced geo-annotation, then infers tracks; tested on four real-world photo datasets it achieves 48%~70% inference accuracy and results in over 80% users exposed with only a few geo-annotated seeds. A countermeasure adapts ring signatures into an AGS prototype that provides adjustable K-anonymity and measurable protection strength.
AI-written summary · about
related talks
15mA Closer Look: Evaluating Location Privacy Empirically
23mAnalysis of Privacy Protections in Fitness Tracking Social Networks -or- You can run, but can you hide?
01h 20mLocation privacy: what problems - which solutions?
14mEstablishing Privacy Metrics for Genomic Data Analysis
39mA Run a Day Won't Keep the Hacker Away: Inference Attacks in Fitness Tracking Social Networks
40mNew Face, Who Dis? Protecting Privacy in a World of Surveillance
46mFrom Health Data to Location Traces: Addressing Spatiotemporal Correlations in Differential Privacy
30mDisambiguating Explicit and Implicit Geographic References...