Encrypted Spiking Neural Networks Based on Adaptive Differential Privacy Mechanism

Entropy, Journal Year: 2025, Volume and Issue: 27(4), P. 333 - 333

Published: March 22, 2025

Spike neural networks (SNNs) perform excellently in various domains. However, SNNs based on differential privacy (DP) protocols introduce uniform noise to the gradient parameters, which may affect trade-off between model efficiency and personal privacy. Therefore, adaptive private SNN (ADPSNN) is proposed this work. It dynamically adjusts budget correlations output spikes labels. In addition, added parameters according budget. The ADPSNN tested four datasets with different spiking neurons including leaky integrated-and-firing (LIF) integrate-and-fire (IF) models. Experimental results show that LIF neuron provides superior utility MNIST (accuracy 99.56%) Fashion-MNIST 92.26%) datasets, while IF performs well CIFAR10 90.67%) CIFAR100 66.10%) datasets. Compared existing methods, accuracy of improved by 0.09% 3.1%. has many potential applications, such as image classification, health care, intelligent driving.

Language: Английский

Robust support vector machine based on the bounded asymmetric least squares loss function and its applications in noise corrupted data

Advanced Engineering Informatics, Journal Year: 2025, Volume and Issue: 65, P. 103371 - 103371

Published: April 29, 2025

Language: Английский

A Noise-Resistant Model for Graph-based Fraud Detection

Information Processing & Management, Journal Year: 2025, Volume and Issue: 62(5), P. 104198 - 104198

Published: May 2, 2025

Language: Английский