Mayfly: Private Aggregate Insights from Ephemeral Streams of On-Device User Data
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866915795125141504 |
|---|---|
| author | Bian, Christopher Cheu, Albert Chiknavaryan, Stanislav Gong, Zoe Gruteser, Marco Guinan, Oliver Guzman, Yannis Kairouz, Peter Lagzdin, Artem McKenna, Ryan Ni, Grace Roth, Edo Spivak, Maya Van Overveldt, Timon Yi, Ren |
| author_facet | Bian, Christopher Cheu, Albert Chiknavaryan, Stanislav Gong, Zoe Gruteser, Marco Guinan, Oliver Guzman, Yannis Kairouz, Peter Lagzdin, Artem McKenna, Ryan Ni, Grace Roth, Edo Spivak, Maya Van Overveldt, Timon Yi, Ren |
| contents | This paper introduces Mayfly, a federated analytics approach enabling aggregate queries over ephemeral on-device data streams without central persistence of sensitive user data. Mayfly minimizes data via on-device windowing and contribution bounding through SQL-programmability, anonymizes user data via streaming differential privacy (DP), and mandates immediate in-memory cross-device aggregation on the server -- ensuring only privatized aggregates are revealed to data analysts. Deployed for a sustainability use case estimating transportation carbon emissions from private location data, Mayfly computed over 4 million statistics across more than 500 million devices with a per-device, per-week DP $\varepsilon = 2$ while meeting strict data utility requirements. To achieve this, we designed a new DP mechanism for Group-By-Sum workloads leveraging statistical properties of location data, with potential applicability to other domains. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_07962 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Mayfly: Private Aggregate Insights from Ephemeral Streams of On-Device User Data Bian, Christopher Cheu, Albert Chiknavaryan, Stanislav Gong, Zoe Gruteser, Marco Guinan, Oliver Guzman, Yannis Kairouz, Peter Lagzdin, Artem McKenna, Ryan Ni, Grace Roth, Edo Spivak, Maya Van Overveldt, Timon Yi, Ren Cryptography and Security Databases H.2.8; K.4.1; H.4 This paper introduces Mayfly, a federated analytics approach enabling aggregate queries over ephemeral on-device data streams without central persistence of sensitive user data. Mayfly minimizes data via on-device windowing and contribution bounding through SQL-programmability, anonymizes user data via streaming differential privacy (DP), and mandates immediate in-memory cross-device aggregation on the server -- ensuring only privatized aggregates are revealed to data analysts. Deployed for a sustainability use case estimating transportation carbon emissions from private location data, Mayfly computed over 4 million statistics across more than 500 million devices with a per-device, per-week DP $\varepsilon = 2$ while meeting strict data utility requirements. To achieve this, we designed a new DP mechanism for Group-By-Sum workloads leveraging statistical properties of location data, with potential applicability to other domains. |
| title | Mayfly: Private Aggregate Insights from Ephemeral Streams of On-Device User Data |
| topic | Cryptography and Security Databases H.2.8; K.4.1; H.4 |
| url | https://arxiv.org/abs/2412.07962 |