Efficient Spiking Point Mamba for Point Cloud Analysis

Bio-inspired Spiking Neural Networks (SNNs) provide an energy-efficient way to extract 3D spatio-temporal features. However, existing 3D SNNs have struggled with long-range dependencies until the recent emergence of Mamba, which offers superior computational efficiency and sequence modeling capability. In this work, we propose Spiking Point Mamba (SPM), the first Mamba-based SNN in the 3D domain. Naively adapting Mamba to 3D SNNs, though, is hindered by temporal dynamics mismatch and spike-induced information loss. Thus, we first introduce Hierarchical Dynamic Encoding (HDE), an improved direct encoding method that effectively introduces dynamic temporal mechanism. Then, we propose Spiking Mamba Block $(S M B)$, which builds upon Mamba while learning inter-time-step features and minimizing information loss caused by spikes. Finally, to further boost performance, we adopt an asymmetric SNN-ANN architecture for spike-based pretraining and finetune. Compared with the previous state-of-the-art SNN models, SPM improves overall accuracy by $\mathbf{+ 6. 2 \%, ~} \mathbf{+ 6. 1 \%}$, and $\mathbf{+ 7. 4 \%}$ on three variants of ScanObjectNN, and boosts instance mIOU by $\mathbf{+ 1. 9 \%}$ on ShapeNetPart. Meanwhile, its energy consumption is at most $\mathbf{1 2. 6} \times$ lower than that of its ANN counterpart. Code: https://github.com/PeppaWu/SPM.