Unveiling the Spatial-temporal Effective Receptive Fields of Spiking Neural Networks

[✒️Openreview] [📄Arxiv Preprint]

This repository provides the official implementation of the paper "Unveiling the Spatial-temporal Effective Receptive Fields of Spiking Neural Networks" (NeurIPS 2025).

📁 File Structure

.
├── erf_compute # Main folder for ST-ERF
├── det # Codebase for Detection Experiments (Originated from mmdetection by OpenMMLab)
├── seg # Codebase for Segmentation Experiments (Originated from mmsegmentation by OpenMMLab)
└── README.md

---

🚀 Quick Start

Environment Setup

Set your python environment with pytorch (lts or 2.6.0+)...

Play with Spatial-temporal Effective Receptive Field

Go to the directory erf_compute/. The structure of this folder is as follows:

.
├── izhikevich.py
├── LICENSE
├── spatial_Erf # Spatial ERF Code
│   ├── erf_scnn # S-ERF for Spiking-CNNs
│   └── erf_sdt  # S-ERF for Spiking-Transformers
└── temporal_erf_compute.py # Temporal ERF Code

There's README.md in these folders, please check it out!

More visual-aids and checkpoints

Go to [🤗ericzhang0328/Spatial-temporal-ERF] and get more you want!

🙏 Acknowledgments

This whole project is influenced by Understanding the Effective Receptive Field in Deep Convolutional Neural Networks . Thanks for their remarkable research!

📄 Citation

If you find this work useful, please cite our paper:

@inproceedings{
zhang2025unveiling,
title={Unveiling the Spatial-temporal Effective Receptive Fields of Spiking Neural Networks},
author={Jieyuan Zhang and Xiaolong Zhou and Shuai Wang and Wenjie Wei and Hanwen Liu and Qian Sun and Malu Zhang and Yang Yang and Haizhou Li},
booktitle={The Thirty-ninth Annual Conference on Neural Information Processing Systems},
year={2025},
url={https://openreview.net/forum?id=tYnJC5ba6j}
}

---

📧 Contact

For questions regarding this implementation, please contact: Jieyuan/Eric 📧 ericzh_uestc@std.uestc.edu.cn