Research map / Federated learning and privacy / Differential privacy
Protecting Geolocation Privacy of Photo Collections
Jinghan Yang, Ayan Chakrabarti, Yevgeniy Vorobeychik
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Replace greedy photo deletion with optimal deletion selection | VGG-16 collection geolocation scores · R-16, top-1 guarantee | fraction of photos deleted: 0.37 -> 0.32 | Improved |
| Replace greedy photo deletion with optimal deletion selection | VGG-16 collection geolocation scores · R-16, top-5 guarantee | fraction of photos deleted: 0.53 -> 0.45 | Improved |
| Replace greedy photo deletion with optimal deletion selection | VGG-16 collection geolocation scores · R-32, top-1 guarantee | fraction of photos deleted: 0.43 -> 0.31 | Improved |
| Replace greedy photo deletion with optimal deletion selection | VGG-16 collection geolocation scores · R-64, top-5 guarantee | fraction of photos deleted: 0.64 -> 0.48 | Improved |
| Replace greedy photo deletion with optimal deletion selection | VGG-16 collection geolocation scores · Albums, top-1 guarantee | fraction of photos deleted: 0.33 -> 0.21 | Improved |
| Replace greedy photo deletion with optimal deletion selection | VGG-16 collection geolocation scores · R-16, 12.5% deletion budget | protected-k: 0.46 -> 0.67 | Improved |
| Replace greedy photo deletion with optimal deletion selection | VGG-16 collection geolocation scores · R-32, 50% deletion budget | protected-k: 7.15 -> 17.49 | Improved |
| Aggregate per-image location scores across a collection instead of using a single image | VGG-16 geolocation classifier · R-16 | top-1 accuracy: 0.20 -> 0.97 | Improved |
| Use a higher proxy-score margin for black-box protection | Black-box top-k guarantee method · R-128 | success rate | Mixed |
Closest papers by meaning
- GEO-Detective: Unveiling Location Privacy Risks in Images with LLM Agents (ECCV 2026 · similarity 0.87)
- GeoShield: Safeguarding Geolocation Privacy from Vision-Language Models via Adversarial Perturbations (AAAI 2026 · similarity 0.87)
- Do Vision-Language Models Respect Contextual Integrity in Location Disclosure? (ICLR 2026 · similarity 0.85)
- GeoClip: Geometry-Aware Clipping for Differentially Private SGD (NeurIPS 2025 · similarity 0.83)
- GeoRanker: Distance-Aware Ranking for Worldwide Image Geolocalization (NeurIPS 2025 · similarity 0.80)
- Faster Algorithms for User-Level Private Stochastic Convex Optimization (NeurIPS 2024 · similarity 0.80)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
