ReverB-SNN: Reversing Bit of the Weight and Activation for Spiking Neural Networks

Yufei Guo, Yuhan Zhang, Zhou Jie, Xiaode Liu, Xin Tong, Yuanpei Chen, Weihang Peng, Zhe Ma
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:21220-21231, 2025.

Abstract

The Spiking Neural Network (SNN), a biologically inspired neural network infrastructure, has garnered significant attention recently. SNNs utilize binary spike activations for efficient information transmission, replacing multiplications with additions, thereby enhancing energy efficiency. However, binary spike activation maps often fail to capture sufficient data information, resulting in reduced accuracy. To address this challenge, we advocate reversing the bit of the weight and activation, called ReverB, inspired by recent findings that highlight greater accuracy degradation from quantizing activations compared to weights. Specifically, our method employs real-valued spike activations alongside binary weights in SNNs. This preserves the event-driven and multiplication-free advantages of standard SNNs while enhancing the information capacity of activations. Additionally, we introduce a trainable factor within binary weights to adaptively learn suitable weight amplitudes during training, thereby increasing network capacity. To maintain efficiency akin to vanilla ReverB, our trainable binary weight SNNs are converted back to standard form using a re-parameterization technique during inference. Extensive experiments across various network architectures and datasets, both static and dynamic, demonstrate that our approach consistently outperforms state-of-the-art methods.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-guo25t, title = {{R}ever{B}-{SNN}: Reversing Bit of the Weight and Activation for Spiking Neural Networks}, author = {Guo, Yufei and Zhang, Yuhan and Jie, Zhou and Liu, Xiaode and Tong, Xin and Chen, Yuanpei and Peng, Weihang and Ma, Zhe}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {21220--21231}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/guo25t/guo25t.pdf}, url = {https://proceedings.mlr.press/v267/guo25t.html}, abstract = {The Spiking Neural Network (SNN), a biologically inspired neural network infrastructure, has garnered significant attention recently. SNNs utilize binary spike activations for efficient information transmission, replacing multiplications with additions, thereby enhancing energy efficiency. However, binary spike activation maps often fail to capture sufficient data information, resulting in reduced accuracy. To address this challenge, we advocate reversing the bit of the weight and activation, called ReverB, inspired by recent findings that highlight greater accuracy degradation from quantizing activations compared to weights. Specifically, our method employs real-valued spike activations alongside binary weights in SNNs. This preserves the event-driven and multiplication-free advantages of standard SNNs while enhancing the information capacity of activations. Additionally, we introduce a trainable factor within binary weights to adaptively learn suitable weight amplitudes during training, thereby increasing network capacity. To maintain efficiency akin to vanilla ReverB, our trainable binary weight SNNs are converted back to standard form using a re-parameterization technique during inference. Extensive experiments across various network architectures and datasets, both static and dynamic, demonstrate that our approach consistently outperforms state-of-the-art methods.} }
Endnote
%0 Conference Paper %T ReverB-SNN: Reversing Bit of the Weight and Activation for Spiking Neural Networks %A Yufei Guo %A Yuhan Zhang %A Zhou Jie %A Xiaode Liu %A Xin Tong %A Yuanpei Chen %A Weihang Peng %A Zhe Ma %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-guo25t %I PMLR %P 21220--21231 %U https://proceedings.mlr.press/v267/guo25t.html %V 267 %X The Spiking Neural Network (SNN), a biologically inspired neural network infrastructure, has garnered significant attention recently. SNNs utilize binary spike activations for efficient information transmission, replacing multiplications with additions, thereby enhancing energy efficiency. However, binary spike activation maps often fail to capture sufficient data information, resulting in reduced accuracy. To address this challenge, we advocate reversing the bit of the weight and activation, called ReverB, inspired by recent findings that highlight greater accuracy degradation from quantizing activations compared to weights. Specifically, our method employs real-valued spike activations alongside binary weights in SNNs. This preserves the event-driven and multiplication-free advantages of standard SNNs while enhancing the information capacity of activations. Additionally, we introduce a trainable factor within binary weights to adaptively learn suitable weight amplitudes during training, thereby increasing network capacity. To maintain efficiency akin to vanilla ReverB, our trainable binary weight SNNs are converted back to standard form using a re-parameterization technique during inference. Extensive experiments across various network architectures and datasets, both static and dynamic, demonstrate that our approach consistently outperforms state-of-the-art methods.
APA
Guo, Y., Zhang, Y., Jie, Z., Liu, X., Tong, X., Chen, Y., Peng, W. & Ma, Z.. (2025). ReverB-SNN: Reversing Bit of the Weight and Activation for Spiking Neural Networks. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:21220-21231 Available from https://proceedings.mlr.press/v267/guo25t.html.

Related Material