arXiv:2605.27412v1 Announce Type: cross 
Abstract: Spiking Neural Networks (SNNs) have emerged with promising energy-efficient property, yet a substantial performance gap persists compared to Artificial Neural Networks (ANNs). This gap stems from at least two key limitations: first, conventional spiking neurons offer limited information representation capacity, underutilizing the rich dynamics of membrane potentials; second, fixed surrogate gradient (SG) functions across time steps leads to imprecise gradient propagation, impeding effective direct training. To address these two challenges, we propose a new direct training algorithm with three core innovations: first, a circulate-firing spiking neuron model that enhances information representation capacity by leveraging membrane potentials more effectively; second, a time-step-wise learnable surrogate gradient function, enabling accurate gradient estimation during backpropagation; third, a positive-negative balanced loss function to achieve equilibrium between positive and negative membrane potentials and further boost SNN performance. Extensive experiments demonstrate that our methods achieve competitive performance across multiple datasets. Our methods can generalize seamlessly to advanced architectures of Transformer, consistently outperforming existing methods. Our work highlights the effectiveness of further harnessing intrinsic membrane dynamics of SNNs for performance improvement, and thus open a new avenue for advancing high-performance spiking neural architectures.

تم اقتراح خوارزمية جديدة للتدريب المباشر لشبكات الأعصاب النابضة (SNN) تتضمن ابتكارات مثل نموذج الخلايا العصبية ذات الإطلاق الدائري، ودالة تدرج بديلة قابلة للتعلم، ودالة خسارة متوازنة. تهدف هذه التطورات إلى تحسين قدرة تمثيل المعلومات وتحسين تقدير التدرج أثناء التدريب، مما يعالج الفجوات الكبيرة في الأداء مقارنة بالشبكات العصبية الاصطناعية (ANN) التقليدية.

Se ha propuesto un nuevo algoritmo de entrenamiento directo para las Redes Neuronales Espigadas (SNN), que presenta innovaciones como un modelo de neurona de disparo circular, una función de gradiente de sustitución aprendible y una función de pérdida equilibrada. Estos avances buscan mejorar la capacidad de representación de la información y la estimación del gradiente durante el entrenamiento, abordando las significativas brechas de rendimiento en comparación con las Redes Neuronales Artificiales (ANN) tradicionales.

Un nouvel algorithme d'entraînement direct pour les réseaux de neurones à impulsion (SNN) a été proposé, présentant des innovations telles qu'un modèle de neurone à tir circulaire, une fonction de gradient de substitution apprenable et une fonction de perte équilibrée. Ces avancées visent à améliorer la capacité de représentation de l'information et à améliorer l'estimation du gradient pendant l'entraînement, abordant des lacunes de performance significatives par rapport aux réseaux de neurones artificiels (ANN) traditionnels.

A new direct training algorithm for Spiking Neural Networks (SNNs) has been proposed, featuring innovations such as a circulate-firing neuron model, a learnable surrogate gradient function, and a balanced loss function. These advancements aim to enhance the information representation capacity and improve gradient estimation during training, addressing significant performance gaps compared to traditional Artificial Neural Networks (ANNs).