SAFformer:Improving Spiking Transformer via Active Predictive Filtering
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917474459451392 |
|---|---|
| author | Xie, Zequan Zeng, Weiming Chen, Yunhua Ling, Sichang Chen, Tongyang Xiao, Jinsheng |
| author_facet | Xie, Zequan Zeng, Weiming Chen, Yunhua Ling, Sichang Chen, Tongyang Xiao, Jinsheng |
| contents | Spiking Neural Networks (SNNs) offer notable advantages in biological plausibility and energy efficiency, making them promising candidates for building low-power Transformers. However, existing Spiking Transformers largely adhere to a passive reactive paradigm, which struggles to focus on task-relevant information and incurs substantial computational overhead when processing redundant visual data. To overcome this fundamental yet underexplored limitation, we propose SAFformer, a novel Spiking Transformer architecture based on an active predictive filtering paradigm. Inspired by the brain's predictive coding mechanism, SAFformer actively suppresses predictable signals and focuses on salient visual features. Extensive experiments show that SAFformer establishes new state-of-the-art performance on CIFAR-10/100 and CIFAR10-DVS. Remarkably, on ImageNet-1K, it achieves 80.50% Top-1 accuracy with only 26.58M parameters and an energy consumption of 5.88 mJ, demonstrating an exceptional balance between accuracy and efficiency. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_08270 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | SAFformer:Improving Spiking Transformer via Active Predictive Filtering Xie, Zequan Zeng, Weiming Chen, Yunhua Ling, Sichang Chen, Tongyang Xiao, Jinsheng Computer Vision and Pattern Recognition Artificial Intelligence 68T07 I.2.10 Spiking Neural Networks (SNNs) offer notable advantages in biological plausibility and energy efficiency, making them promising candidates for building low-power Transformers. However, existing Spiking Transformers largely adhere to a passive reactive paradigm, which struggles to focus on task-relevant information and incurs substantial computational overhead when processing redundant visual data. To overcome this fundamental yet underexplored limitation, we propose SAFformer, a novel Spiking Transformer architecture based on an active predictive filtering paradigm. Inspired by the brain's predictive coding mechanism, SAFformer actively suppresses predictable signals and focuses on salient visual features. Extensive experiments show that SAFformer establishes new state-of-the-art performance on CIFAR-10/100 and CIFAR10-DVS. Remarkably, on ImageNet-1K, it achieves 80.50% Top-1 accuracy with only 26.58M parameters and an energy consumption of 5.88 mJ, demonstrating an exceptional balance between accuracy and efficiency. |
| title | SAFformer:Improving Spiking Transformer via Active Predictive Filtering |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence 68T07 I.2.10 |
| url | https://arxiv.org/abs/2605.08270 |