Private machine learning classification based on fully homomorphic encryption¶
Authors: Xiaoqiang Sun; Peng Zhang; Joseph K. Liu; Jianping Yu; Weixin Xie
Year: 2018
Venue: IEEE Transactions on Emerging Topics in Computing
type: scheme collection: ppml status: auto
Links: DOI
Summary¶
Auto-extracted summary (unreviewed)
Proposes an improved BGV-style FHE scheme built on HElib that applies relinearization and modulus switching selectively (skipping them when the next operation on a product ciphertext is additive or absent) to reduce ciphertext size and noise. Using additive and multiplicative homomorphism it implements a homomorphic comparison protocol whose interaction count drops from 3 to 1, plus private hyperplane decision-based, Naive Bayes and decision tree classification; simulations show the scheme and the private decision tree classifier are more efficient than two prior schemes.
Abstract¶
Machine learning classification is an useful tool for trend prediction by analyzing big data. As supporting homomorphic operations over encrypted data without decryption, fully homomorphic encryption (FHE) contributes to machine learning classification without leaking user privacy, especially in the outsouring scenario. In this paper, we propose an improved FHE scheme based on HElib, which is a FHE library implemented based on Brakerski’s FHE scheme. Our improvement focuses on two aspects. On the one hand, we first use the relinearization technique to reduce the ciphertext size, and then the modulus switching technique is used to reduce the modulus and decryption noise. On the other hand, we need no relinearization and modulus switching if there is additive homomorphic or no homomorphic operation in the multiplicative ciphertext’s next homomorphic operation. Homomorphic comparison protocol, private hyperplane decisionbased classification and private Na€Ive Bayes classification are implemented by additive homomorphic and multiplicative homomorphic first. In our homomorphic comparison protocol, the number of interactions is reduced from 3 to 1. We choose the proposed FHE scheme to implement private decision tree classification. Simulation results show that the efficiency of our FHE scheme and implementation of private decision tree classification are more efficient than other two schemes.
Source: pdf
Extracted details¶
| Field | Value | Source | Confidence |
|---|---|---|---|
| contribution | Improved HElib-based FHE scheme with selective relinearization and modulus switching | llm | 0.65 |
| contribution | Homomorphic comparison protocol with interactions reduced from 3 to 1 | llm | 0.65 |
| contribution | Private hyperplane decision-based, Naive Bayes and decision tree classification over encrypted data | llm | 0.65 |
| schemes | BGV | llm | 0.65 |
| libraries | HElib | llm | 0.65 |
| models | hyperplane decision-based classifier | llm | 0.65 |
| models | Naive Bayes | llm | 0.65 |
| models | decision tree | llm | 0.65 |
| tasks | private classification in outsourced computation | llm | 0.65 |
| threat model | outsourced (cloud) classification without leaking user data; interactive comparison protocol | llm | 0.65 |
| interaction | interactive (HE+MPC) | llm | 0.65 |
| code url | https://github.com/shaih/HElib (URL found in the full text) | regex | 0.50 |
| llm relevance | low | llm | 0.65 |
| key result | interaction rounds of homomorphic comparison: 1 (reduced from 3) (homomorphic comparison protocol) | llm | 0.65 |
| key result | efficiency: more efficient than two other schemes (simulation of FHE scheme and private decision tree classification) | llm | 0.65 |
Keyword mentions¶
Counts of vocabulary terms in the full text (first page of occurrence in brackets).
- scheme: Paillier 8 [p.1], Leveled / somewhat HE 4 [p.2]
- library: HElib 38 [p.1]
- model: Decision tree 28 [p.1], SVM 1 [p.1], k-NN 1 [p.12], Naive Bayes 1 [p.12]
- dataset: Breast cancer (Wisconsin) 1 [p.9], UCI repository 1 [p.9]
- hardware: GPU 1 [p.9]
- technique: SIMD packing 74 [p.1], Relinearization 14 [p.1], Modulus switching 14 [p.1], Bootstrapping 3 [p.12]
- assumption: RLWE 11 [p.2], LWE 8 [p.2], Lattice-based 7 [p.2]
- task: Medical / healthcare 8 [p.1]
- topic: Homomorphic encryption 148 [p.1], Open source 2 [p.12]
Files¶
| Path (under papers/) | Collection | Pages | Size | SHA-256 | Note |
|---|---|---|---|---|---|
references/ppml/inference/sun-2020_private-ml-classification-based-on-fhe.pdf |
ppml/inference | 13 | 1.4 MB | facae4cb7a15 |
was Sun et al. - 2020 - Private Machine Learning Classification Based on Fully Homomorphic Encryption.pdf |
Database key: private-machine-learning-classification-based-on-fully-homomorphic · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set private-machine-learning-classification-based-on-fully-homomorphic COLUMN VALUE