EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

SAD-SNN: Spatial-Activation Distillation for High-Performance Spiking Neural Networks

Chongxiao Qu iD, Qian Zhang iD, Chenxiao Dou iD, Xiaohu Li iD, Xinyu Chen iD, Zhenyu Zhao, Baoqing Zeng iD, Xinwei Yao iD

Sensors · 02.08.2026 · 26 (15) · 4877

DOI: 10.3390/s26154877 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Sensors, online erschienen am 02.08.2026. Verfasst von Chongxiao Qu und 7 weiteren Autor:innen. Beteiligte Einrichtung: School of Electronic Science and Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China. Die Arbeit nennt 49 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

A Spiking Neural Network (SNN) is a kind of brain-inspired and event-driven network, which is becoming a promising energy-efficient alternative to Artificial Neural Networks (ANNs). In recent years, SNN methods have been successfully applied in the fields of electromagnetic signal processing and image signal processing, particularly in application scenarios that require low energy consumption. However, the performance of SNNs by direct training is far from satisfactory. In this paper, we study a novel learning method named SAD-SNN (Spatial-Activation Distillation for Spiking Neural Networks), which utilizes the ANN model to guide the SNN model learning. Unlike prior works that rely on element-wise feature alignment approaches, SAD-SNN aligns spatial-activation maps at different resolutions of the teacher and student networks. Specifically, we introduce a direct alignment approach, which defines a spatial-activation loss and normalizes the representation vectors of ANN and SNN, to alleviate the unexpected precision loss. This enables the knowledge of teacher ANNs to be effectively transferred to train student SNNs. On three image classification datasets, our proposed SAD-SNN outperforms other SNN training methods no matter whether homogeneous or heterogeneous teacher ANNs are used. Furthermore, we apply SAD-SNN to the electromagnetic signal detection task, demonstrating strong generalization ability and superior performance. In conclusion, the experimental results on various tasks and SNN architectures demonstrate that our method is a general and effective solution that significantly improves the learning of student SNNs with only two time steps.

Schlagwörter

Stichwörter der Autor:innen

electromagnetic signal detectionimage classificationknowledge distillationspiking neural networks

Autor:innen und Institutionen

  • School of Electronic Science and Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China.
  • China Nanhu Academy of Electronics and Information Technology, Jiaxing 314002, China.
  • Academy of Advanced Interdisciplinary Science and Technology, Zhejiang University of Technology, Hangzhou 310014, China.

ORCID-Profile: Chongxiao Qu, Qian Zhang, Chenxiao Dou, Xiaohu Li, Xinyu Chen, Baoqing Zeng, Xinwei Yao

Zitieren

APA

Qu, C., Zhang, Q., Dou, C., Li, X., Chen, X., Zhao, Z., Zeng, B., & Yao, X. (2026). SAD-SNN: Spatial-Activation Distillation for High-Performance Spiking Neural Networks. Sensors, 26(15), 4877. https://doi.org/10.3390/s26154877

Vancouver

Qu C, Zhang Q, Dou C, Li X, Chen X, Zhao Z, et al. SAD-SNN: Spatial-Activation Distillation for High-Performance Spiking Neural Networks. Sensors. 2026;26(15):4877. doi:10.3390/s26154877

BibTeX
@article{Qu2026,
  author = {Qu, Chongxiao and Zhang, Qian and Dou, Chenxiao and Li, Xiaohu and Chen, Xinyu and Zhao, Zhenyu and Zeng, Baoqing and Yao, Xinwei},
  title = {SAD-SNN: Spatial-Activation Distillation for High-Performance Spiking Neural Networks},
  journal = {Sensors},
  year = {2026},
  volume = {26},
  number = {15},
  pages = {4877},
  doi = {10.3390/s26154877}
}
Kostenlos starten →