Temporal Misalignment in ANN-SNN Conversion and Its Mitigation via Probabilistic Spiking Neurons

Fuente: arXiv
Guardado en:
Detalles Bibliográficos
Autores principales: Bojković, Velibor, Wu, Xiaofeng, Gu, Bin
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866916623950020608
author Bojković, Velibor
Wu, Xiaofeng
Gu, Bin
author_facet Bojković, Velibor
Wu, Xiaofeng
Gu, Bin
contents Spiking Neural Networks (SNNs) offer a more energy-efficient alternative to Artificial Neural Networks (ANNs) by mimicking biological neural principles, establishing them as a promising approach to mitigate the increasing energy demands of large-scale neural models. However, fully harnessing the capabilities of SNNs remains challenging due to their discrete signal processing and temporal dynamics. ANN-SNN conversion has emerged as a practical approach, enabling SNNs to achieve competitive performance on complex machine learning tasks. In this work, we identify a phenomenon in the ANN-SNN conversion framework, termed temporal misalignment, in which random spike rearrangement across SNN layers leads to performance improvements. Based on this observation, we introduce biologically plausible two-phase probabilistic (TPP) spiking neurons, further enhancing the conversion process. We demonstrate the advantages of our proposed method both theoretically and empirically through comprehensive experiments on CIFAR-10/100, CIFAR10-DVS, and ImageNet across a variety of architectures, achieving state-of-the-art results.
format Preprint
id arxiv_https___arxiv_org_abs_2502_14487
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Temporal Misalignment in ANN-SNN Conversion and Its Mitigation via Probabilistic Spiking Neurons
Bojković, Velibor
Wu, Xiaofeng
Gu, Bin
Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
Spiking Neural Networks (SNNs) offer a more energy-efficient alternative to Artificial Neural Networks (ANNs) by mimicking biological neural principles, establishing them as a promising approach to mitigate the increasing energy demands of large-scale neural models. However, fully harnessing the capabilities of SNNs remains challenging due to their discrete signal processing and temporal dynamics. ANN-SNN conversion has emerged as a practical approach, enabling SNNs to achieve competitive performance on complex machine learning tasks. In this work, we identify a phenomenon in the ANN-SNN conversion framework, termed temporal misalignment, in which random spike rearrangement across SNN layers leads to performance improvements. Based on this observation, we introduce biologically plausible two-phase probabilistic (TPP) spiking neurons, further enhancing the conversion process. We demonstrate the advantages of our proposed method both theoretically and empirically through comprehensive experiments on CIFAR-10/100, CIFAR10-DVS, and ImageNet across a variety of architectures, achieving state-of-the-art results.
title Temporal Misalignment in ANN-SNN Conversion and Its Mitigation via Probabilistic Spiking Neurons
topic Machine Learning
Artificial Intelligence
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2502.14487