让脉冲神经网络在点云任务中仅更新5%参数，还能保持高效能。
Parameter-Efficient Fine-Tuning for Spiking Point Cloud Models
- 通过调节膜电位衰减和阈值实现神经元内在自适应。
- 在ModelNet40上达92.4%准确率，扫描对象数据集上85.6%。
- 适合资源受限设备上的低功耗点云模型轻量微调。
脉冲神经网络（SNNs）通过事件驱动计算为资源受限设备上的点云分析提供了低功耗解决方案。然而，现有的预训练脉冲点云模型依赖全量微调进行下游任务适配，带来显著的参数与存储开销。此外，二值脉冲传播会抑制任务相关的亚阈值信息。为此，我们提出SpikePEFT，首个面向脉冲点云模型的参数高效微调框架。具体而言，内在动态调优（IDT）自适应调节膜电位衰减与放电阈值，实现神经元内在高效适应，同时保持预训练突触变换冻结；沉默状态消歧适配（SSDA）从有信息量的沉默状态中恢复任务相关特征，为下游适配提供更丰富证据。多基准测试表明SpikePEFT有效且高效：在ModelNet40上达到92.4%准确率，在最具挑战性的ScanObjectNN(PB_T50_RS)分类划分上达85.6%，仅更新约5%的可训练参数，并维持SNN的能效优势。本工作为类脑视觉模型的参数高效适配提供了可行路径。
原文摘要 · Abstract (English)
Spiking Neural Networks (SNNs) offer energy-efficient solutions for point cloud analysis on resource-constrained devices through event-driven computation. However, existing pre-trained spiking point cloud models rely on full fine-tuning for downstream task adaptation, incurring substantial parameter and storage overhead. Furthermore, binary spike propagation suppresses task-relevant sub-threshold information. To address these issues, we propose SpikePEFT, the first parameter-efficient fine-tuning framework for spiking point cloud models. Specifically, Intrinsic Dynamics Tuning (IDT) adaptively modulates membrane decay and firing thresholds, enabling efficient neuron-intrinsic adaptation while keeping the pre-trained synaptic transformations frozen. Moreover, Silent-State Disambiguation Adaptation (SSDA) recovers task-relevant information from informative silent states, thereby providing richer evidence for downstream adaptation. Extensive experiments across multiple benchmarks demonstrate the effectiveness and efficiency of SpikePEFT. In particular, our method achieves 92.4% accuracy on ModelNet40 and 85.6\% on the most challenging classification split ScanObjectNN(PB\_T50\_RS) while updating only about 5% of the trainable parameters and preserving the energy efficiency of SNNs. This work provides a promising step toward parameter-efficient adaptation of neuromorphic vision models.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。