Pang, Qi ; Zhu, Jinhao ; Möllering, Helen ; Zheng, Wenting ; Schneider, Thomas (2024)
BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers.
45th IEEE Symposium on Security and Privacy (IEEE S&P'24). San Francisco, USA (20.05.2024 - 23.05.2024)
doi: 10.1109/SP54263.2024.00130
Konferenzveröffentlichung, Bibliographie
Kurzbeschreibung (Abstract)
The advent of transformers has brought about significant advancements in traditional machine learning tasks. However, their pervasive deployment has raised concerns about the potential leakage of sensitive information during inference. Existing approaches using secure multiparty computation (MPC) face limitations when applied to transformers due to the extensive model size and resource-intensive matrix-matrix multiplications. In this paper, we present BOLT, a privacy-preserving inference framework for transformer models that supports efficient matrix multiplications and nonlinear computations. Combined with our novel machine learning optimizations, BOLT reduces the communication cost by 10.91×. Our evaluation on diverse datasets demonstrates that BOLT maintains comparable accuracy to floating-point models and achieves 4.8-9.5× faster inference across various network settings compared to the state-of-the-art system.
| Typ des Eintrags: | Konferenzveröffentlichung |
|---|---|
| Erschienen: | 2024 |
| Autor(en): | Pang, Qi ; Zhu, Jinhao ; Möllering, Helen ; Zheng, Wenting ; Schneider, Thomas |
| Art des Eintrags: | Bibliographie |
| Titel: | BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers |
| Sprache: | Englisch |
| Publikationsjahr: | Mai 2024 |
| Verlag: | IEEE |
| Buchtitel: | Proceedings: 45th IEEE Symposium on Security and Privacy: SP 2024 |
| Veranstaltungstitel: | 45th IEEE Symposium on Security and Privacy (IEEE S&P'24) |
| Veranstaltungsort: | San Francisco, USA |
| Veranstaltungsdatum: | 20.05.2024 - 23.05.2024 |
| DOI: | 10.1109/SP54263.2024.00130 |
| Zugehörige Links: | |
| Kurzbeschreibung (Abstract): | The advent of transformers has brought about significant advancements in traditional machine learning tasks. However, their pervasive deployment has raised concerns about the potential leakage of sensitive information during inference. Existing approaches using secure multiparty computation (MPC) face limitations when applied to transformers due to the extensive model size and resource-intensive matrix-matrix multiplications. In this paper, we present BOLT, a privacy-preserving inference framework for transformer models that supports efficient matrix multiplications and nonlinear computations. Combined with our novel machine learning optimizations, BOLT reduces the communication cost by 10.91×. Our evaluation on diverse datasets demonstrates that BOLT maintains comparable accuracy to floating-point models and achieves 4.8-9.5× faster inference across various network settings compared to the state-of-the-art system. |
| Fachbereich(e)/-gebiet(e): | 20 Fachbereich Informatik 20 Fachbereich Informatik > Praktische Kryptographie und Privatheit DFG-Sonderforschungsbereiche (inkl. Transregio) DFG-Sonderforschungsbereiche (inkl. Transregio) > Sonderforschungsbereiche DFG-Graduiertenkollegs DFG-Graduiertenkollegs > Graduiertenkolleg 2050 Privacy and Trust for Mobile Users DFG-Sonderforschungsbereiche (inkl. Transregio) > Sonderforschungsbereiche > SFB 1119: CROSSING – Kryptographiebasierte Sicherheitslösungen als Grundlage für Vertrauen in heutigen und zukünftigen IT-Systemen |
| Hinterlegungsdatum: | 25 Jul 2024 07:30 |
| Letzte Änderung: | 25 Jul 2024 07:30 |
| PPN: | |
| Export: | |
| Suche nach Titel in: | TUfind oder in Google |
 |
Frage zum Eintrag |
Optionen (nur für Redakteure)
 |
Redaktionelle Details anzeigen |
Drucken
Impressum