Noise-Injected Spiking Graph Convolution for Energy-Efficient 3D Point Cloud Denoising

2025-02-27

AI Summary

This work pioneers the application of Spiking Neural Networks (SNNs) to 3D point cloud denoising—a challenging regression task—balancing high accuracy with ultra-low energy consumption. To this end, we propose a noise-injected Spiking Graph Convolution mechanism that enables perturbation-aware spike-based representation learning. We design two architectures: a fully spiking variant and an ANN-SNN hybrid, both integrated with an event-driven graph signal processing strategy. Evaluated on PU-Net and PC-Net benchmarks, our method achieves denoising performance comparable to full-precision ANNs within only a few time steps, while drastically reducing energy consumption. This breakthrough overcomes longstanding accuracy and efficiency bottlenecks of SNNs in 3D regression tasks, establishing a practical pathway for deploying brain-inspired chips and enabling energy-efficient 3D sensing systems.

Technology Category

• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Machine Learning: Neuro-Symbolic Learning
• Computer Vision: 3D Computer Vision

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding
• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments

Abstract

Spiking neural networks (SNNs), inspired by the spiking computation paradigm of the biological neural systems, have exhibited superior energy efficiency in 2D classification tasks over traditional artificial neural networks (ANNs). However, the regression potential of SNNs has not been well explored, especially in 3D point cloud processing.In this paper, we propose noise-injected spiking graph convolutional networks to leverage the full regression potential of SNNs in 3D point cloud denoising. Specifically, we first emulate the noise-injected neuronal dynamics to build noise-injected spiking neurons. On this basis, we design noise-injected spiking graph convolution for promoting disturbance-aware spiking representation learning on 3D points. Starting from the spiking graph convolution, we build two SNN-based denoising networks. One is a purely spiking graph convolutional network, which achieves low accuracy loss compared with some ANN-based alternatives, while resulting in significantly reduced energy consumption on two benchmark datasets, PU-Net and PC-Net. The other is a hybrid architecture that combines ANN-based learning with a high performance-efficiency trade-off in just a few time steps. Our work lights up SNN's potential for 3D point cloud denoising, injecting new perspectives of exploring the deployment on neuromorphic chips while paving the way for developing energy-efficient 3D data acquisition devices.

Problem

• Explores SNNs for 3D point cloud denoising
• Reduces energy consumption in denoising tasks
• Combines ANN and SNN for efficient processing

Innovation

• Noise-injected spiking neurons
• Spiking graph convolution
• Hybrid SNN-ANN architecture