Phase-dependent plasticity: A brain-inspired unsupervised learning algorithm for deep neural networks

Sijia Lu

doi:10.1016/j.neucom.2026.132707

Phase-dependent plasticity: A brain-inspired unsupervised learning algorithm for deep neural networks

2026-01-16

0

PRE

AI

DOI：10.1016/j.neucom.2026.132707

原文链接

原文求助

摘要

En 中文

• Integrate the cross-domain mechanisms of SNNs and ANNs. This paper integrates biologically inspired mechanisms from Spiking Neural Networks (SNNs) into the architecture of Artificial Neural Networks (ANNs), providing a promising approach for unsupervised deep learning; • Propose a novel Frequency-Phase Coding (FPC) mechanism. The proposed FPC mechanism synthesizes frequency-based encoding from SNNs with real-valued encoding of ANNs, introducing a hybrid time-domain and real-valued representation to improve computational efficiency; • Introduce a Phase-Dependent Neuron (PDN) model and Phase-Dependent Plasticity (PDP) learning rule. The PDN model establishes a connection between Linear Leaky-Integrate-and-Fire (LIF) neurons and ReLU-based units in ANNs, while the PDP learning rule updates synaptic weights, thereby enhancing the capacity for unsupervised learning in deep neural networks.

Phase-dependent plasticity: A brain-inspired unsupervised learning algorithm for deep neural networks

摘要

期刊

机构

引用论文