SpyKing—Privacy-preserving framework for Spiking Neural Networks¶
Authors: Farzad Nikfam; Alberto Marchisio; Maurizio Martina; Muhammad Shafique
Year: 2025
Venue: Frontiers in Neuroscience
type: system collection: ppml status: auto
Links: DOI
Summary¶
Auto-extracted summary (unreviewed)
SpyKing is a framework for running spiking neural networks (SNNs) and conventional DNNs on FHE-encrypted data, used here to compare the two model families under encryption while sweeping encryption parameters such as the plaintext modulus t. Experiments on MNIST, FashionMNIST and CIFAR-10 show that FHE greatly increases computational cost but remains viable in accuracy, and that SNNs reach up to 35% higher absolute accuracy than DNNs on encrypted data when t is small.
Abstract¶
Artificial intelligence (AI) models, frequently built using deep neural networks (DNNs), have become integral to many aspects of modern life. However, the vast amount of data they process is not always secure, posing potential risks to privacy and safety. Fully Homomorphic Encryption (FHE) enables computations on encrypted data while preserving its confidentiality, making it a promising approach for privacy-preserving AI. This study evaluates the performance of FHE when applied to DNNs and compares it with Spiking Neural Networks (SNNs), which more closely resemble biological neurons and, under certain conditions, may achieve superior results. Using the SpyKing framework, we analyze key challenges in encrypted neural computations, particularly the limitations of FHE in handling non-linear operations. To ensure a comprehensive evaluation, we conducted experiments on the MNIST, FashionMNIST, and CIFAR10 datasets while systematically varying encryption parameters to optimize SNN performance. Our results show that FHE significantly increases computational costs but remains viable in terms of accuracy and data security. Furthermore, SNNs achieved up to 35% higher absolute accuracy than DNNs on encrypted data with low values of the plaintext modulus t . These findings highlight the potential of SNNs in privacy-preserving AI and underscore the growing need for secure yet efficient neural computing solutions.
Source: crossref
Extracted details¶
| Field | Value | Source | Confidence |
|---|---|---|---|
| contribution | SpyKing framework for FHE-based inference with spiking neural networks | llm | 0.60 |
| contribution | systematic comparison of DNNs and SNNs under FHE constraints | llm | 0.60 |
| contribution | analysis of FHE limitations for non-linear operations in encrypted neural computation | llm | 0.60 |
| contribution | sweep of encryption parameters (plaintext modulus t) to optimize encrypted SNN accuracy | llm | 0.60 |
| schemes | FHE | llm | 0.60 |
| models | LeNet-5 | llm | 0.60 |
| models | SNN | llm | 0.60 |
| models | DNN | llm | 0.60 |
| datasets | MNIST | llm | 0.60 |
| datasets | FashionMNIST | llm | 0.60 |
| datasets | CIFAR-10 | llm | 0.60 |
| tasks | image classification | llm | 0.60 |
| tasks | encrypted spiking neural network inference | llm | 0.60 |
| interaction | non-interactive FHE | llm | 0.60 |
| limitations | FHE significantly increases computational cost | llm | 0.60 |
| limitations | FHE has difficulty handling non-linear operations | llm | 0.60 |
| llm relevance | low | llm | 0.60 |
| key result | accuracy gain of SNN over DNN on encrypted data: up to 35% absolute (low plaintext modulus t; MNIST / FashionMNIST / CIFAR-10) | llm | 0.60 |
Keyword mentions¶
Counts of vocabulary terms in the full text (first page of occurrence in brackets).
- library: Pyfhel 1 [p.20]
- model: Spiking neural network 22 [p.1], LeNet 7 [p.1]
- dataset: MNIST 2 [p.1], Fashion-MNIST 2 [p.1], CIFAR-10 2 [p.1]
- assumption: Lattice-based 2 [p.20]
- task: Medical / healthcare 2 [p.1]
- topic: Homomorphic encryption 19 [p.1]
Files¶
| Path (under papers/) | Collection | Pages | Size | SHA-256 | Note |
|---|---|---|---|---|---|
references/ppml/inference/nikfam-2025_spyking-privacy-preserving-framework-for-spiking-neural-networks.pdf |
ppml/inference | 20 | 674.0 KB | 8246d2f6a7d9 |
was Nikfam et al. - 2025 - SpyKing—Privacy-preserving framework for Spiking Neural Networks.pdf |
Database key: spykingprivacy-preserving-framework-for-spiking-neural-networks · created 2026-11-05 · updated 2026-11-05 · edit with python3 tools/paperdb.py set spykingprivacy-preserving-framework-for-spiking-neural-networks COLUMN VALUE