Spatial–Temporal Spiking Feature Pruning in Spiking Transformer
Lists
Only you see this
Abstract
Spiking neural networks (SNNs) are known for brain-inspired architecture and low power consumption. Leveraging biocompatibility and self-attention mechanism, Spiking Transformers become the most promising SNN architecture with high accuracy. However, Spiking Transformers still faces the challenge of high training costs, such as a 51$M$ network requiring 181 training hours on ImageNet. In this work, we explore feature pruning to reduce training costs and overcome two challenges: high pruning ratio and lightweight pruning methods. We first analyze the spiking features and find the potential for a high pruning ratio. The majority of information is concentrated on a part of the spiking features in spiking transformer, which suggests that we can keep this part of the tokens and prune the others. To achieve lightweight, a parameter-free spatial–temporal spiking feature pruning method is proposed, which uses only a simple addition-sorting operation. The spiking features/tokens with high spike accumulation values are selected for training. The others are pruned and merged through a compensation module called Softmatch. Experimental results demonstrate that our method reduces training costs without compromising image classification accuracy. On ImageNet, our approach reduces the training time from 181 to 128 h while achieving comparable accuracy (83.13% versus 83.07%).
BibTeX
@article{zhou2025spatialtemporal,
author = {Zhaokun Zhou and Kaiwei Che and Jun Niu and Man Yao and Guoqi Li and Li Yuan and Guibo Luo and Yuesheng Zhu},
title = {{Spatial–Temporal Spiking Feature Pruning in Spiking Transformer}},
journal = {IEEE Transactions on Cognitive and Developmental Systems},
year = {2025},
url = {https://openreview.net/forum?id=WqPlLFMHkD},
}Loading related content...