Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Privacy-Preserving Spiking Neural Networks: A Deep Dive into Encryption Parameter Optimisation

Mahitha Pulivathi, Ana Fontes Rodrigues, Isibor Kennedy Ihianle, Andreas Oikonomou, Srinivas Boppu, Pedro Machado

TL;DR

This work introduces BioEncryptSNN, a spiking-neural-network framework that performs privacy-preserving encryption by converting ciphertext into spike trains and leveraging temporal dynamics for encryption and decryption. It combines an encryption framework with a bio-inspired neuron model and a multi-layer architecture (Input, Hidden, Output, plus Lateral Inhibition, Noise, and Teaching layers) to encode, process, and reconstruct data, guided by temporal metrics such as Purpura Distance and Van Rossum distance. Empirical results benchmark BioEncryptSNN against AES, DES, and RSA, showing up to 4.1x faster encryption/decryption than PyCryptodome AES in some configurations and demonstrating noise resilience and data integrity under encrypted processing. The study highlights practical implications for energy-efficient, privacy-preserving computation and outlines future directions including encoding optimisations, homomorphic encryption integrations, and broader secure-application deployments.

Abstract

Deep learning is widely applied to modern problems through neural networks, but the growing computational and energy demands of these models have driven interest in more efficient approaches. Spiking Neural Networks (SNNs), the third generation of neural networks, mimic the brain's event-driven behaviour, offering improved performance and reduced power use. At the same time, concerns about data privacy during cloud-based model execution have led to the adoption of cryptographic methods. This article introduces BioEncryptSNN, a spiking neural network based encryption-decryption framework for secure and noise-resilient data protection. Unlike conventional algorithms, BioEncryptSNN converts ciphertext into spike trains and exploits temporal neural dynamics to model encryption and decryption, optimising parameters such as key length, spike timing, and synaptic connectivity. Benchmarked against AES-128, RSA-2048, and DES, BioEncryptSNN preserved data integrity while achieving up to 4.1x faster encryption and decryption than PyCryptodome's AES implementation. The framework demonstrates scalability and adaptability across symmetric and asymmetric ciphers, positioning SNNs as a promising direction for secure, energy-efficient computing.

Privacy-Preserving Spiking Neural Networks: A Deep Dive into Encryption Parameter Optimisation

TL;DR

This work introduces BioEncryptSNN, a spiking-neural-network framework that performs privacy-preserving encryption by converting ciphertext into spike trains and leveraging temporal dynamics for encryption and decryption. It combines an encryption framework with a bio-inspired neuron model and a multi-layer architecture (Input, Hidden, Output, plus Lateral Inhibition, Noise, and Teaching layers) to encode, process, and reconstruct data, guided by temporal metrics such as Purpura Distance and Van Rossum distance. Empirical results benchmark BioEncryptSNN against AES, DES, and RSA, showing up to 4.1x faster encryption/decryption than PyCryptodome AES in some configurations and demonstrating noise resilience and data integrity under encrypted processing. The study highlights practical implications for energy-efficient, privacy-preserving computation and outlines future directions including encoding optimisations, homomorphic encryption integrations, and broader secure-application deployments.

Abstract

Deep learning is widely applied to modern problems through neural networks, but the growing computational and energy demands of these models have driven interest in more efficient approaches. Spiking Neural Networks (SNNs), the third generation of neural networks, mimic the brain's event-driven behaviour, offering improved performance and reduced power use. At the same time, concerns about data privacy during cloud-based model execution have led to the adoption of cryptographic methods. This article introduces BioEncryptSNN, a spiking neural network based encryption-decryption framework for secure and noise-resilient data protection. Unlike conventional algorithms, BioEncryptSNN converts ciphertext into spike trains and exploits temporal neural dynamics to model encryption and decryption, optimising parameters such as key length, spike timing, and synaptic connectivity. Benchmarked against AES-128, RSA-2048, and DES, BioEncryptSNN preserved data integrity while achieving up to 4.1x faster encryption and decryption than PyCryptodome's AES implementation. The framework demonstrates scalability and adaptability across symmetric and asymmetric ciphers, positioning SNNs as a promising direction for secure, energy-efficient computing.
Paper Structure (19 sections, 12 equations, 7 figures, 9 tables)

This paper contains 19 sections, 12 equations, 7 figures, 9 tables.

Table of Contents

  1. Introduction
  2. Related works
  3. Methodology
  4. Encryption Framework
  5. Neuron Model
  6. Neuron Creation and Input Layer (Layer 1)
  7. Hidden Layer (Layer 2)
  8. Output Layer (Layer 3)
  9. Lateral Inhibition Layer
  10. Noise Layer
  11. Teaching Layer
  12. Evaluation metrics
  13. Network Training and Testing
  14. Purpura Distance
  15. Van Rossum Metric
  16. ...and 4 more sections

Figures (7)

  • Figure 1: A Bio-Inspired Framework for secure Encryption
  • Figure 2: Biological neuron and its association with an artificial spiking neuron
  • Figure 3: Weight change in neurons based on the learning
  • Figure 4: BioEncryptSNN Encryption Architecture
  • Figure 5: BioEncryptSNN Decryption Architecture
  • ...and 2 more figures