Fully Homomorphic Encryption with Applications to Privacy-Preserving Machine Learning¶
Year: 2023
type: thesis collection: ppml status: auto
Summary¶
Auto-extracted summary (unreviewed)
This Harvard College bachelor's thesis (Michael Gul, 2023) is an expository treatment of fully homomorphic encryption written for readers without a cryptography background. It builds up the mathematical foundations (lattice problems, LWE and Ring-LWE), describes two popular FHE schemes including GSW, and surveys how FHE is currently applied to machine learning, emphasizing settings such as healthcare where privacy regulations otherwise block ML. Its stated contribution is the completeness of the exposition rather than new results.
Extracted details¶
| Field | Value | Source | Confidence |
|---|---|---|---|
| contribution | Self-contained exposition of FHE from first principles | llm | 0.60 |
| contribution | Presentation of LWE and Ring-LWE foundations and lattice-based encryption | llm | 0.60 |
| contribution | Description of two popular FHE schemes (including GSW) | llm | 0.60 |
| contribution | Survey of current uses of FHE for privacy-preserving machine learning | llm | 0.60 |
| schemes | LWE | llm | 0.60 |
| schemes | Ring-LWE | llm | 0.60 |
| schemes | GSW | llm | 0.60 |
| tasks | privacy-preserving machine learning | llm | 0.60 |
| tasks | healthcare ML (motivation) | llm | 0.60 |
| threat model | user sends encrypted data to a remote ML service provider | llm | 0.60 |
| interaction | non-interactive FHE | llm | 0.60 |
| code url | https://github.com/harvard-ml-courses/cs181-textbook/blob/master/Textbook.pdf (URL found in the full text) | regex | 0.50 |
| llm relevance | low | llm | 0.60 |
Keyword mentions¶
Counts of vocabulary terms in the full text (first page of occurrence in brackets).
- scheme: CKKS 42 [p.4], GSW 28 [p.4], Multi-key HE 11 [p.5], Leveled / somewhat HE 10 [p.7], Partially homomorphic encryption 6 [p.30], TFHE 5 [p.33], FHEW 3 [p.33], BFV 2 [p.33]
- model: CNN 7 [p.54]
- dataset: MNIST 3 [p.57], Credit card / German credit 1 [p.14], ECG 1 [p.51]
- technique: Bootstrapping 23 [p.4], Federated learning 17 [p.5], Rescaling 4 [p.46], Softmax 4 [p.53], ReLU 4 [p.53], Convolution 4 [p.54], Sigmoid 2 [p.53]
- assumption: LWE 76 [p.4], RLWE 49 [p.4], Lattice-based 46 [p.4]
- task: Medical / healthcare 17 [p.3], Language modeling / generation 6 [p.18], Finance 4 [p.8], Genomics 3 [p.51], Text classification 2 [p.51]
- topic: Homomorphic encryption 129 [p.3]
Files¶
| Path (under papers/) | Collection | Pages | Size | SHA-256 | Note |
|---|---|---|---|---|---|
references/ppml/theses-and-textbooks/gul-2023_fhe-with-applications-to-privacy-preserving-ml-bsc-thesis.pdf |
ppml/theses-and-textbooks | 69 | 857.0 KB | 27930874ff62 |
was Gul - Fully Homomorphic Encryption with Applications to Privacy-Preserving Machine Learning.pdf |
Database key: fully-homomorphic-encryption-with-applications-to-privacy-preserving · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set fully-homomorphic-encryption-with-applications-to-privacy-preserving COLUMN VALUE