Towards Green Wearable Computing: A Physics-Aware Spiking Neural Network for Energy-Efficient IMU-ba... Activation: 物理感知, spiking, physics-aware, snn
Wearable IMU-based Human Activity Recognition (HAR) relies heavily on Deep Neural Networks (DNNs), which are burdened by immense computational and buffering demands. Their power-hungry floating-point operations and rigid requirement to process complete temporal windows severely cripple battery-constrained edge devices. While Spiking Neural Networks (SNNs) offer extreme event-driven energy efficiency, standard architectures struggle with complex biomechanical topologies and temporal gradient degr
Evaluated across seven diverse datasets, PAS-Net achieves state-of-the-art accuracy while replacing dense operations with sparse 0.1 pJ integer accumulations.
# 核心架构示例
# 脉冲神经网络示例
import torch
import torch.nn as nn
class SNNLayer(nn.Module):
def __init__(self, input_size, hidden_size):
super().__init__()
self.fc = nn.Linear(input_size, hidden_size)
self.threshold = 1.0
def forward(self, x):
# 前向传播与脉冲生成
mem = self.fc(x)
spike = (mem > self.threshold).float()
return spike
技能自动生成于: 2026-04-15 来源: arXiv自动化研究工作流