Controlling Noise Budget of Fully Homomorphic Encryption in Secure Machine Learning
Guardado en:
| Publicado en: | PQDT - Global (2025) |
|---|---|
| Autor principal: | |
| Publicado: |
ProQuest Dissertations & Theses
|
| Materias: | |
| Acceso en línea: | Citation/Abstract Full Text - PDF |
| Etiquetas: |
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
MARC
| LEADER | 00000nab a2200000uu 4500 | ||
|---|---|---|---|
| 001 | 3215574598 | ||
| 003 | UK-CbPIL | ||
| 020 | |a 9798315783046 | ||
| 035 | |a 3215574598 | ||
| 045 | 2 | |b d20250101 |b d20251231 | |
| 084 | |a 189128 |2 nlm | ||
| 100 | 1 | |a Mahmood, Fatima | |
| 245 | 1 | |a Controlling Noise Budget of Fully Homomorphic Encryption in Secure Machine Learning | |
| 260 | |b ProQuest Dissertations & Theses |c 2025 | ||
| 513 | |a Dissertation/Thesis | ||
| 520 | 3 | |a Fully Homomorphic Encryption (FHE) offers a promising path to privacy preserving machine learning, but its widespread adoption is constrained by the challenge of noise budget management. This thesis analyzes the control of noise growth in encrypted logistic regression models using the CKKS encryption scheme via the HEAAN library, within the iDASH2017 framework. A two level parameter tuning methodology was developed, modifying internal encryption settings and adjusting runtime parameters to explore over 720 unique configurations. The experiment was automated through a Python based pipeline that included command execution, result logging, CSV merging, and 3D visualization. Three algorithms were developed: a noise budget simulation tool, a log retrieval mechanism, and a parameter suggestion engine. These tools were applied across multiple datasets to systematically identify optimal configurations that maximize noise budget utilization without causing execution failure. The results revealed key trends linking encryption parameters to model accuracy and computational stability, with parameter adjustments improving both the Area Under the Curve (AUC) and execution depth. The findings contribute a practical framework for executing encrypted machine learning tasks while maintaining both data confidentiality and operational feasibility. The developed heuristics and evaluation strategies serve as a foundation for future research in adaptive, efficient, and scalable privacy-preserving machine learning. | |
| 653 | |a Computer science | ||
| 653 | |a Computer engineering | ||
| 653 | |a Acoustics | ||
| 773 | 0 | |t PQDT - Global |g (2025) | |
| 786 | 0 | |d ProQuest |t ProQuest Dissertations & Theses Global | |
| 856 | 4 | 1 | |3 Citation/Abstract |u https://www.proquest.com/docview/3215574598/abstract/embedded/7BTGNMKEMPT1V9Z2?source=fedsrch |
| 856 | 4 | 0 | |3 Full Text - PDF |u https://www.proquest.com/docview/3215574598/fulltextPDF/embedded/7BTGNMKEMPT1V9Z2?source=fedsrch |