Towards Accurate Binary Spiking Neural Networks: Learning with Adaptive Gradient Modulation Mechanism (AAAI 2025 Oral)

Yu Liang, Wenjie Wei, Ammar Belatreche, Honglin Cao, Zijian Zhou, Shuai Wang, Malu Zhang, Yang Yang

University of Electronic Science and Technology of China, Northumbria University

Abstract

Binary Spiking Neural Networks (BSNNs) inherit the eventdriven paradigm of SNNs, while also adopting the reduced storage burden of binarization techniques. These distinct advantages grant BSNNs lightweight and energy-effcient characteristics, rendering them ideal for deployment on resourceconstrained edge devices. However, due to the binary synaptic weights and non-differentiable spike function, effectively training BSNNs remains an open question. In this paper, we conduct an in-depth analysis of the challenge for BSNN learning, namely the frequent weight sign fipping problem. To mitigate this issue, we propose an Adaptive Gradient Modulation Mechanism (AGMM), which is designed to reduce the frequency of weight sign fipping by adaptively adjusting the gradients during the learning process. The proposed AGMM can enable BSNNs to achieve faster convergence speed and higher accuracy, effectively narrowing the gap between BSNNs and their full-precision equivalents. We validate AGMM on both static and neuromorphic datasets, and results indicate that it achieves state-of-the-art results among BSNNs. This work substantially reduces storage demands and enhances SNNs’ inherent energy effciency, making them highly feasible for resource-constrained environments.

You can use the train_cifar.sh file to train agmm-bsnn.

./train_cifar.sh

Citation

If you find this project useful in your research, please consider cite:

@inproceedings{liang2025towards,
  title={Towards Accurate Binary Spiking Neural Networks: Learning with Adaptive Gradient Modulation Mechanism},
  author={Liang, Yu and Wei, Wenjie and Belatreche, Ammar and Cao, Honglin and Zhou, Zijian and Wang, Shuai and Zhang, Malu and Yang, Yang},
  booktitle={Proceedings of the AAAI Conference on Artificial Intelligence},
  volume={39},
  number={2},
  pages={1402--1410},
  year={2025}
}